Open tcllib tickets: Reproduced

Triage of all 509 open and Pending tickets of tcllib (bugs, patches and feature requests) on 2026-11-15. Each ticket was checked against tcllib trunk on Linux with Tcl 9.1, 9.0, 8.6 and 8.5, using local servers for the network modules and a Windows 11 VM for Windows-only tickets. Outdated, superseded, duplicate and misfiled tickets are listed separately. Rows marked patch link to a prototype fix with a regression test (147 in all, in tcllib-triage/prototypes). Updated 2026-11-16.

Reproduced 135

🔧d3d41178032026practcl initstubs does not cover Tcl 9patchREPRODUCED (9.1, 8.6). Generated loader code has Tcl_InitStubs(interp, "8.6", 0), which fails in Tcl 9 (package require Tcl 8.6 -> version conf…

Open ticket d3d4117803 · Bug · (unused) · filed 2026-09-14

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in af1fe948f6 on practcl-initstubs-tcl9, not merged: Fix d3d4117803: practcl generated Tcl_InitStubs(interp, "8.6", 0) and Tk_InitStubs with "8.6", which fails on Tcl/Tk 9. Require "8.6-" instead.
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch practcl-initstubs-tcl9: as Peter suggested, the generated Tcl_InitStubs and Tk_InitStubs calls require "8.6-" instead of "8.6", in practcl.tcl and in its build source. Test practcl-loader-001.
Full verdict

REPRODUCED (9.1, 8.6). Generated loader code has Tcl_InitStubs(interp, "8.6", 0), which fails in Tcl 9 (package require Tcl 8.6 -> version conflict). Tk_InitStubs is the same.

  • Module practcl 0.16.6
  • Tested practcl::module ... generate-loader output on 9.1 and 8.6. package require Tcl 8.6 in tclsh9.1 gives "version conflict".
  • Cause practcl.tcl:generate-loader (and its source build/class/module.tcl) hardcodes "8.6"
  • Prototype prototypes/d3d4117803.patch (73-line patch, +20/-4: "8.6" -> "8.6-" for Tcl and Tk in practcl.tcl and build/class/module.tcl; test practcl-loader-001 fails without the fix)
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed. The generated init code requires "8.6", which fails in Tcl 9. A proposed fix uses "8.6-" for Tcl and Tk, with a test.
⬜f6624cf52b2026dns::resolve leaks memory on errorpatchREPRODUCED (9.1, 8.6): a bad option, a missing nameserver, missing udp support or missing tls leave a ::dns::N array behind. Proposed fix in bran…

Open ticket f6624cf52b · Bug · dns · filed 2026-09-04

Full verdict

REPRODUCED (9.1, 8.6): a bad option, a missing nameserver, missing udp support or missing tls leave a ::dns::N array behind. Proposed fix in branch bug-f6624cf52ba4 (f8ed54f73b), not merged. That branch has a bug: in the bad-option path it unsets state before building the option list, so the error ends with "must be " and nothing after it.

  • Module dns 1.6.1
  • Tested tclsh9.1/8.6 repro; the branch diff applied to the checkout; dns tests
  • Cause dns.tcl ::dns::resolve error returns do not unset state
  • Prototype prototypes/f6624cf52b.patch (55 lines, tests dns-3.1–3.4). Same as the branch, with the option list built before the unset, and the original package require tls error message kept.
  • Suggested action fix the branch (move unset state after set opts), add the tests, merge
  • Suggested ticket comment The branch fixes the leak, but in the bad-option case it unsets state before listing the options, so the message ends with "must be ". Suggest moving the unset one line down. Tests are in the triage patch.
🔧05349414362026Tcllib - ::mime::parsetype (mime.tcl:1017) - state(content) is never setpatchREPRODUCED (9.1, 8.6), when ::errorCode/::errorInfo do not exist. Normally hidden because something during package require mime creates them, as …

Open ticket 0534941436 · Bug · (unused) · filed 2026-08-16

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in b1d77474a2 on mime-catch-options, not merged: Fix 0534941436: mime read errorCode and errorInfo after a successful catch, so a stale value made the first parse of a multipart message return text/plain. Us
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch mime-catch-options: all places that read errorCode/errorInfo after catch now use the catch options dict, so a stale or missing global value no longer makes the first multipart parse return text/plain. Tests mime-15.0 and mime-15.1.
Full verdict

REPRODUCED (9.1, 8.6), when ::errorCode/::errorInfo do not exist. Normally hidden because something during package require mime creates them, as the reporter says.

  • Module mime 1.7.2
  • Tested reporter's reproducer with unset ::errorCode ::errorInfo: the first parse gives text/plain, the second gives multipart/mixed, on both 9.1 and 8.6
  • Cause mime.tcl reads $errorCode/$errorInfo after catch even on success, in initializeaux, parsetype, getbody, copymessage, buildmessage, parseaddress and addr_next. parsepart swallows the resulting error.
  • Prototype prototypes/0534941436.patch (373-line patch, +51/-54; uses catch ... result opts / return -options $opts $result and drops global errorCode errorInfo. Tests mime-15.0 (multipart parse) and mime-15.1 (parseaddress/buildmessage/copymessage/getbody) fail without the fix. The full mime suite passes on 9.1 and 8.6.)
  • Suggested action apply prototype. Separately, parsepart silently ignores Content-Type parse errors (reporter's hardening suggestion).
  • Suggested ticket comment Confirmed on trunk when ::errorCode/::errorInfo don't exist. A proposed fix replaces the errorCode/errorInfo reads with catch options dicts at all ten sites, with tests.
🔧0f65bfa2d32026::struct::graph::op::tarjan infinite looppatchREPRODUCED (9.1, 8.6). This is a regression of Bug 2815302.

Open ticket 0f65bfa2d3 · Bug · struct :: graph · filed 2026-08-08

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 1ad56464d4 on graph-tarjan-node-spaces, not merged: Fix 0f65bfa2d3: struct::graph::op::tarjan hung and isCutVertex? gave wrong answers for node names containing spaces. The fix for Bug 2815302 (746fd7f965) was
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch graph-tarjan-node-spaces: the change in the attached diff (struct::set exclude instead of subtract in TarjanSub), plus the same fix in isCutVertex?, which still uses subtract on trunk. This is a regression of Bug 2815302: the fix 746fd7f965 was undone by the GSoC merge 42c21b6c75. Tests tarjan-3.9 and isCutVertex?-3.4.
Full verdict

REPRODUCED (9.1, 8.6). This is a regression of Bug 2815302.

  • Module struct::graph::op 0.11.4 (graphops.tcl)
  • Tested a node named {beamsect_C C} with an arc makes tarjan hang (killed by timeout). isCutVertex? gives wrong results for such names (1 instead of 0).
  • Cause graphops.tcl TarjanSub and isCutVertex? use struct::set subtract (set operand) instead of exclude (single element). This was fixed in 746fd7f965 (2009-07-10) and reintroduced by the GSoC integration 42c21b6c75 (2009-09-15). Fleury still has the fix.
  • Prototype prototypes/0f65bfa2d3.patch (81 lines, tests tarjan-3.9 and isCutVertex?-3.4). It restores exclude in both places. graphops.test passes 434/434 on 9.1/8.6. Without the fix, tarjan-3.9 hangs and isCutVertex?-3.4 fails. The reporter's attached diff is not in the local repo (attachment not synced), so the prototype was written independently.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: a regression of Bug 2815302 (the subtract→exclude fix was lost in 2009-09). isCutVertex? is affected too. Proposed fix with tests tarjan-3.9 and isCutVertex?-3.4.
🔧523f7e05562026Incorrect versions/files provided compared to pkgIndex.tclpatchREPRODUCED (9.1, 8.6). All five mismatches are still on trunk.

Open ticket 523f7e0556 · Bug · (unused) · filed 2026-08-07

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in d828c0f28f on pkgindex-version-fixes, not merged: Fix 61e49798ee, 523f7e0556, cdbdd399e4: pkgIndex.tcl entries that do not match the code. nettool's index sourced nexttool.tcl, so package require nettool
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch pkgindex-version-fixes: fixes the entries for rest, doctools::toc, math::filters, tcl::chan::std and nettool, and adds tests that compare each pkgIndex.tcl entry with what the code provides.
Full verdict

REPRODUCED (9.1, 8.6). All five mismatches are still on trunk.

  • Module rest 1.8, doctools::toc 2.1, math::filters 0.4, nettool 0.5.4, tcl::chan::std 1.2
  • Tested script that package require -exacts every tcllib package from the source tree in a fresh child interp, on 9.1 and 8.6. Got the "provided instead" errors for doctools::toc, math::filters, tcl::chan::std, and nexttool.tcl not found for nettool (and udpcluster, which requires it). rest's index says 1.7 while rest.tcl provides 1.8 (bumped in 0b3dd048f8 without the index). Extra finding: struct 1.5 (struct1.tcl) requires struct::matrix 1.2.1, which no longer exists.
  • Cause pkgIndex.tcl not updated with version bumps. The nettool typo came in with 94a3829047 (2024-09-13) and is in the 2.0 release.
  • Prototype prototypes/523f7e0556.patch (267-line patch, +125/-6: 5 pkgIndex.tcl fixes, rest.man VERSION 1.8, a pkgIndex-vs-provide test in container.test, filtergen.test, nettool.test, and new rest/rest.test and virtchannel_base/std.test; tests *-pkgindex-1.0 fail without the fix)
  • Suggested action apply prototype. The nettool part duplicates open ticket 61e49798ee (close both). struct 1.5 needs a separate decision.
  • Suggested ticket comment All five still mismatch on trunk. A proposed fix updates the indexes, with tests comparing pkgIndex.tcl against package provide. The nettool typo is also 61e49798ee.
🔧29ff5c3abd2026autoproxy stopped working with tcl-tls on Tcl-9.xpatchREPRODUCED (9.1, 9.0; works on 8.6)

Open ticket 29ff5c3abd · Bug · http :: autoproxy · filed 2026-07-29

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in f9dc4e7b85 on autoproxy-tls-socket-tcl9, not merged: Fix 29ff5c3abd: autoproxy::tls_socket did not work with Tcl 9. http 2.10 calls it at global level, so its upvar of host and state saw the wrong variables. Tak
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch autoproxy-tls-socket-tcl9. http 2.10 calls the registered socket command from http::OpenSocket at global level, so the upvar of host and state in autoproxy::tls_socket sees the wrong variables. The branch takes the target from state(socketinfo). It still depends on http internals, as discussed above; a proper interface in http would be better. autoproxytls.test now runs the tls tests with a …
Full verdict

REPRODUCED (9.1, 9.0; works on 8.6)

  • Module autoproxy 1.8.1
  • Tested autoproxytls.test with the tls package simulated: on 9.1 and 9.0, can't read "uhost" with and without a proxy. Passes on 8.6.
  • Cause autoproxy.tcl tls_socket uses upvar host/upvar state, assuming that http::geturl calls it directly. http 2.10 calls the socket command from http::OpenSocket via namespace eval ::, so the upvars point to global variables. This also means tls_status was being written into a global ::state.
  • Prototype prototypes/29ff5c3abd.patch (153 lines; tls_socket looks up the caller frame http::geturl or http::OpenSocket, takes the target from state(socketinfo), and does nothing if neither is found). For the test: autoproxytls.test now simulates the tls package when it is missing, so the existing tests autoproxy-tls-1.1.0/1.1.1/1.1.3 run everywhere. 1.1.0 and 1.1.1 fail on 9.1 without the fix. With the fix the module passes on 9.1, 9.0, 8.6 and 8.5. The second test named 1.1.1 ("with proxy and http proxy config") stays limited to have_tls: it fails with Tcl 8.6 http (Tcl 2d9b31130b), and on 9.x it would need http::register ... {} 0 1.
  • Suggested action apply prototype
  • Suggested ticket comment http 2.10 calls the socket command from http::OpenSocket via namespace eval ::, so tls_socket's upvar no longer finds host/port. Proposed fix: tls_socket finds the http state array and takes the target from state(socketinfo). The tls tests now run without a real tls package.
⬜f44bcbf0682026Procedure mime::encodingasciiP always returns 0 if the line contains \rpatchREPRODUCED (9.1, 8.6): two bugs

Open ticket f44bcbf068 · Bug · mime · filed 2026-07-29

Full verdict

REPRODUCED (9.1, 8.6): two bugs

  • Module mime 1.7.2
  • Tested mime::encodingasciiP "Simple line\r" → 0. A parsed message (mime::initialize -string of a full mail, no CTE header) gets a computed "Content-Transfer-Encoding: quoted-printable" (base64 with the reporter's charset=Cp1252) from buildmessage, but the body is copied unencoded. That is the garbled mail the reporter saw.
  • Cause (1) encodingasciiP compares $r == {[string length $line] - 1}, a literal string. (2) mime.tcl:getTransferEncoding (new in 1.7) computes [encoding] for non-canonical (parsed) parts too, but copymessageaux only encodes canonical bodies. In 1.6 only the received CTE was written for parsed parts.
  • Prototype prototypes/f44bcbf068.patch (95 lines; fixes the expr, and computes the encoding only for canonical parts; tests mime-16.1 (encodingasciiP), 16.2 (no invented CTE for a parsed body), 16.3 (received CTE kept and decoded); 16.1/16.2 fail without the fix on 9.1 and 8.6)
  • Suggested action apply prototype (applies together with the dcf333e549 and 67ace2b948 patches; all three together: 115/115 pass on 9.1 and 8.6). The reporter's last comment says encodinglineP also forces encoding for long lines. That is correct for canonical bodies, and harmless once parsed bodies no longer get a computed CTE.
  • Suggested ticket comment Confirmed. Besides the encodingasciiP expression, the real problem is that 1.7 adds a computed Content-Transfer-Encoding to parsed messages without encoding the body. Proposed fix: compute it only for canonical parts, with tests.
⬜67ace2b9482026The version 1.7.2 of the mime package creates extra \r characters in the mail bodypatchREPRODUCED (9.1, 8.6), regression in 1.7 (1.6.2 buildmessage built the string by hand, with no channel translation)

Open ticket 67ace2b948 · Bug · mime · filed 2026-07-29

Full verdict

REPRODUCED (9.1, 8.6), regression in 1.7 (1.6.2 buildmessage built the string by hand, with no channel translation)

  • Module mime 1.7.2
  • Tested a parsed message with CRLF body → buildmessage gives "line1\r\r\n"; same for a canonical -encoding 8bit body containing CRLF
  • Cause mime.tcl:buildmessageaux writes through a memchan with -translation crlf; copymessageaux writes unconverted bodies (string and file) as is, so an existing \r\n becomes \r\r\n
  • Prototype prototypes/67ace2b948.patch (125 lines; when the body is unconverted and the output channel translates to crlf, map \r\n → \n first, carrying a trailing \r across 32 KB file chunks; tests mime-17.1 (parsed string), 17.2 (canonical mixed LF/CRLF), 17.3 (parsed -file body with \r at the chunk boundary); all fail without the fix, 17.3 also fails without the carry)
  • Suggested action apply prototype (after dcf333e549/f44bcbf068 is fine)
  • Suggested ticket comment Confirmed: buildmessage writes through a CRLF-translating channel, so CRLF in the body became CR CR LF. Proposed fix with tests: normalize the body's CRLF first when the channel translates.
⬜90cb1c13682026::math::geometry::findLineSegmentIntersection: suddenly no intersection?patchREPRODUCED (9.1, 8.6)

Open ticket 90cb1c1368 · Bug · math :: geometry · filed 2026-07-02

Full verdict

REPRODUCED (9.1, 8.6)

  • Module math, math::geometry 1.4.2
  • Tested findLineSegmentIntersection {0 0 10 0} {5 1e-8 15 1e-8} and {0 0 10 0} {10 1e-8 20 1e-8} print the message to stderr and return none (9.1, 8.6).
  • Cause geometry.tcl. lineSegmentsIntersect treats segments closer than 1e-7 as intersecting, but findLineIntersection returns "none" for exactly parallel, non-collinear lines, so findLineSegmentIntersection reaches the "should never happen" branch and prints to stderr. The adjacency check in the coincident branch also uses exact equality, which ignores the same margin.
  • Prototype prototypes/90cb1c1368.patch (80 lines, test geometry-10.8). The "none" case is merged into the "coincident" handling, the endpoint adjacency check uses the same 1e-7 distance margin, and the stderr output is removed. The test fails without the fix (stderr output, wrong results); geometry.test passes 211/211 on 9.1 and 8.6.
  • Suggested action apply prototype
  • Suggested ticket comment Happens for parallel segments closer than the 1e-7 tolerance of lineSegmentsIntersect. A library should not print to stderr. Proposed fix in prototype.
🔧61e49798ee2026typo in pkgIndex.tclpatchREPRODUCED (9.1, 8.6). Also in the tcllib 2.0 release.

Open ticket 61e49798ee · Bug · nettool · filed 2026-05-22

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in d828c0f28f on pkgindex-version-fixes, not merged: Fix 61e49798ee, 523f7e0556, cdbdd399e4: pkgIndex.tcl entries that do not match the code. nettool's index sourced nexttool.tcl, so package require nettool
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch pkgindex-version-fixes: the index now sources nettool.tcl, as suggested in the ticket. The typo came with 94a3829047 and is in the tcllib 2.0 release, so package require nettool always fails there. The branch also fixes the other index mismatches from 523f7e0556, with tests that compare each index with package provide.
Full verdict

REPRODUCED (9.1, 8.6). Also in the tcllib 2.0 release.

  • Module nettool 0.5.4
  • Tested package require nettool from the source tree gives couldn't read file ".../nettool/nexttool.tcl" on 9.1 and 8.6. The tcllib-2-0 tag has the same pkgIndex.tcl.
  • Cause modules/nettool/pkgIndex.tcl. The typo nexttool.tcl came in with 94a3829047 (2024-09-13, "Tcl 9 portability fixes"). The test suite does not catch it because nettool.test sources nettool.tcl directly. Side note: build/build.tcl, which regenerates pkgIndex, still writes the stale versions 0.5.2 / available_ports 0.1, and ::nettool::version says 0.5.3.
  • Prototype prototypes/61e49798ee.patch (41 lines, test pkgindex-0001). It fixes the file name and adds a test that sources pkgIndex.tcl in a child interp and requires the package. The test fails without the fix. 26/26 pass on 9.1 and 8.6.
  • Suggested action apply the prototype. This one is urgent, because nettool cannot be loaded at all from an installed tcllib 2.0.
  • Suggested ticket comment Confirmed: nettool cannot be loaded from tcllib 2.0 because of this typo (since 94a3829047). Fix with a pkgIndex load test is ready.
⬜a23348be0b2026::yaml::_parsePlainScalarInBlock only parses the first 5 lines of a plain scalarpatchREPRODUCED (9.1, 8.6)

Open ticket a23348be0b · Bug · yaml · filed 2026-05-22

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested the ticket example drops "six". Removing the limit alone still fails beyond about 1000 lines (too many nested evaluations) because of the recursion.
  • Cause yaml.tcl ::yaml::_parsePlainScalarInBlock: leftover if {$loop == 5} return from fc59390525 (2009), plus one recursion level per line.
  • Prototype prototypes/a23348be0b.patch (99 lines, tests yaml-30.1–30.3). Rewrites the recursion as a loop with the same progress check; a 2000-line scalar works.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed; the 5-line cap is a debug leftover. Proposed fix turns the recursion into a loop (no line limit); tests yaml-30.*.
⬜815408482c2026yaml::yaml2dict misidentifies * as alias indicator on plain scalar continuation linespatchREPRODUCED (9.1, 8.6)

Open ticket 815408482c · Bug · yaml · filed 2026-04-12

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested the ticket example gives the anchor error. "key: some long text\n - text\nother: val\n" and "a: x\n y" also lose the continuation line.
  • Cause yaml.tcl ::yaml::_parseBlockNode passes the scalar's own start column as base to _parsePlainScalarInBlock, so continuation lines indented less than the first word end the scalar. The spec only requires them to be more indented than the parent node. The * is then parsed as an alias.
  • Prototype prototypes/815408482c.patch (42 lines, tests yaml-31.1, yaml-31.2). Passes $indent + 1 as the base. Results match PyYAML.
  • Suggested action apply prototype
  • Suggested ticket comment Continuation lines were required to be indented at least to the scalar's first column. Proposed fix uses the parent indentation; tests yaml-31.*.
⬜4dd68ce4e72026::math::statistics::linear-model does not return correct t valuepatchREPRODUCED (9.1, 8.6); fix exists on branch tkt-4dd68ce4e76841f (02df8c0ed9, 2026-03-25), reviewed OK by Harald, not merged

Open ticket 4dd68ce4e7 · Bug · math :: statistics · filed 2026-03-25

Full verdict

REPRODUCED (9.1, 8.6); fix exists on branch tkt-4dd68ce4e76841f (02df8c0ed9, 2026-03-25), reviewed OK by Harald, not merged

  • Module math::statistics 1.6.1
  • Tested reporter's data on trunk: slope t = 21.648 (wrong). With the branch fix: 4.96644 (R/Excel agree). Independently checked in Python: intercept t = 1.07819, slope t = 4.96644, no-intercept slope t = 16.6222. The branch fix gives all three. statistics.test 154/154 → 157/157 on 9.1 and 8.6.
  • Cause math/statistics.tcl:linear-model multiplies $A/$seA and $B/$seB by an extra sqrt(df-2) (sqrt(df-1) without intercept); seA/seB are already standard errors.
  • Prototype prototypes/4dd68ce4e7.patch (82 lines: the branch's fix + test linear-model-1.1, plus linear-model-1.2 (intercept t) and 1.3 (no-intercept t), which I added; all 3 fail without the fix)
  • Suggested action merge the branch (optionally with the two extra tests). Separately, the 5th element "Number of degrees of freedom df" is actually the number of points n (21 here, but df is 19). The reporter's p-value helper passes that to cdf-students-t, so callers get a slightly wrong p. Fix the docs or the value (that changes the API).
  • Suggested ticket comment Confirmed. The branch fix is right: it matches R for the slope, and independent computations for the intercept t and the no-intercept case. Note that the returned "df" is really the number of points, not n-2.
⬜201a0a0bf52026The fileutil::magic compiler does not work with current "magic" files.REPRODUCED (9.1, 8.6)

Open ticket 201a0a0bf5 · Bug · fileutil :: magic · filed 2026-03-16

Full verdict

REPRODUCED (9.1, 8.6)

  • Module fumagic, fileutil::magic::cfront 1.3.3 / filetype (trunk)
  • Tested file-5.46 Magdir (FreeBSD contrib copy, closest available offline). ::fileutil::magic::cfront::generate compressed 0 -- Magdir → about 90 rejected lines ("bad modifier" 56, "unknown type" 14 e.g. guid, "type not implemented" 12, "bad indirect offset type" 8) plus about 295 warnings. Then installed the generated script and ran filetype /bin/ls: it returns the same nonsense NAXOS CAD match as in the report, and some inputs throw can't read "typematch(1)". The shipped 5.34-based recognizer gives the correct ELF answer.
  • Cause cfront's parser/cgen lag the magic(5) syntax added after file-5.34: new types (guid, …), new modifiers and indirect offsets. The bad /bin/ls answer is probably a regex/string test that was parsed wrongly but still emitted.
  • Prototype none (large compiler update, not a small fix)
  • Suggested action keep open. Side note: cfront::compile opens the make magic file with r+ and rewrites it in place, so it modifies the user's input tree. That should be fixed too.
  • Suggested ticket comment Reproduced with the file-5.46 Magdir on Tcl 9.1 and 8.6: about 90 rejected lines (unknown types such as guid, unsupported modifiers and indirect offsets), and the generated recognizer misidentifies /bin/ls. cfront needs updating for the magic(5) syntax added after 5.34.
🔧1ae6b4775e2026zipfile::encode does not preserve file mtimespatchREPRODUCED (9.1, 8.6; existing knownBug test zipfile-encode-encode-6.3)

Open ticket 1ae6b4775e · Bug · zip · filed 2026-03-08

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 410d398079 on zipfile-encode-mtime, not merged: Fix 1ae6b4775e: zipfile::encode stored the ctime of files instead of the modification time. Remove the knownBug constraint from zipfile-encode-encode-6.3.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch zipfile-encode-mtime: use the mtime instead of the ctime. Test zipfile-encode-encode-6.3 is no longer a knownBug.
Full verdict

REPRODUCED (9.1, 8.6; existing knownBug test zipfile-encode-encode-6.3)

  • Module zipfile::encode 0.5.2
  • Tested encode.test with the knownBug constraint removed. 6.3 fails on 9.1 and 8.6 without the fix and passes with it. The whole zip suite passes (33/33) on both.
  • Cause encode.tcl method file: stores $s(ctime) (inode change time) instead of $s(mtime) for the zip timestamp.
  • Prototype prototypes/1ae6b4775e.patch (6 changed lines, test zipfile-encode-encode-6.3 has its knownBug constraint removed). Both files and directories now use mtime.
  • Suggested action apply prototype.
  • Suggested ticket comment file: recorded the ctime instead of the mtime. Proposed fix: use mtime. Test 6.3 is no longer knownBug.
⬜f2b25fff772026Trivial bug in ::csv::split2matrix?patchREPRODUCED (9.1, 8.6): split2matrix -alternate m line sep expand fails with "wrong#args", because the 5-argument branch is labelled 4.

Open ticket f2b25fff77 · Bug · csv · filed 2026-02-04

Full verdict

REPRODUCED (9.1, 8.6): split2matrix -alternate m line sep expand fails with "wrong#args", because the 5-argument branch is labelled 4.

  • Module csv 0.10
  • Tested tclsh9.1, tclsh8.6; csv.test
  • Cause csv.tcl ::csv::split2matrix has a duplicate 4 switch arm. read2matrix has the correct 5.
  • Prototype prototypes/f2b25fff77.patch (19 lines, tests csv-200.4/200.5). A one-character fix, without the submitter's lassign refactor.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed. The 5-argument form fell into the error branch. Fix: label the arm 5.
🔧69d33816bb2026yencode ignores the -file option and can't read using -filepatchREPRODUCED (9.1, 8.6). Three bugs:

Open ticket 69d33816bb · Bug · base64 · filed 2026-02-03

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 7053d9a68b on yencode-ydecode-fixes, not merged: Fix 69d33816bb: ydecode accepted only all-digit file names, so it nearly always returned default.bin; ydecode -file read in text mode (an error on Tcl 9, corr
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch yencode-ydecode-fixes: ydecode takes the name from the rest of the =ybegin line, instead of accepting only digits; -file is opened in binary mode; the bad-option error message no longer fails. Tests yencode-6.0 to 6.2.
Full verdict

REPRODUCED (9.1, 8.6). Three bugs:

  • Module yencode 1.1.4 (modules/base64/yencode.tcl)
    • ydecode parses the name with name=(\d+), so it always returns default.bin unless the name is all digits.
    • ydecode -file opens the file in text mode. On 9.x this throws on 8-bit data; on 8.6 utf-8 sequences in the data are silently corrupted.
    • The bad-option error tries to use the opts array as a scalar ($opts).
  • Tested ticket script and own repro on 9.1/8.6; new tests yencode-6.0/6.1/6.2 fail on both without the fix and pass with it. The rest of the base64 module is unchanged (97 pass, 9 skipped).
  • Cause yencode.tcl ydecode: regexp {name=(\d+)}, open $opts(filename) r, must be -$opts
  • Prototype prototypes/69d33816bb.patch (78 lines, tests yencode-6.0, yencode-6.1, yencode-6.2)
  • Suggested action apply prototype, close as Fixed
  • Suggested ticket comment ydecode only accepted all-digit names, and it read -file in text mode. Both are fixed: the name is taken from the rest of the =ybegin line, and the file is opened in binary mode.
⬜dcf333e5492026Content-Transfer-Encoding header produced twicepatchREPRODUCED (9.1, 8.6), regression in 1.7.x

Open ticket dcf333e549 · Bug · mime · filed 2026-02-02

Full verdict

REPRODUCED (9.1, 8.6), regression in 1.7.x

  • Module mime 1.7.2
  • Tested -encoding 8bit → two CTE lines. Also happens without -encoding whenever an encoding is computed (e.g. application/octet-stream → base64), including inside multipart.
  • Cause 7c25c9a457 (2021) made mime::getheader emit Content-Transfer-Encoding; mime.tcl:copymessageaux prints all getheader headers and then still prints the CTE line itself.
  • Prototype prototypes/dcf333e549.patch (80 lines; removes the second puts; tests mime-15.1–15.7 (no -encoding, 8bit, 7bit, base64, quoted-printable, computed base64, copymessage); 6 fail without the fix on 9.1 and 8.6)
  • Suggested action apply prototype; close 04b8e4fb3c as its duplicate
  • Suggested ticket comment Confirmed. Since 1.7, getheader already includes the header and copymessageaux writes it again. Proposed fix with tests attached; 04b8e4fb3c is the same issue.
⬜a028a78b5b2026Backslashes inside emphasis disable emphasispatchREPRODUCED (9.1, 8.6)

Open ticket a028a78b5b · Bug · markdown · filed 2026-01-24

Full verdict

REPRODUCED (9.1, 8.6)

  • Module markdown, Markdown 1.2.5
  • Tested *a\b* gives <p>*a\b*</p>, and *a\b\** gives <p>*a\b**</p> (9.1, 8.6). Torsten's branch bug-markdown-a028a78b5b7e (33e0f416bb, 2026-01-30) adds only tests and no fix. Some of its expected results (*a\b* giving <em>ab</em>, \c giving c) contradict CommonMark, where a backslash before a non-punctuation character is kept.
  • Cause markdown.tcl:parse_inline, the emphasis regexp (?:[^X\\]|\\X)*. A backslash is allowed only before the delimiter X.
  • Prototype prototypes/a028a78b5b.patch (58 lines, tests basic-2.1..2.4). It changes the regexp to (?:[^X\\]|\\.)*, so any escape pair is consumed and parse_inline then handles the escape. The tests cover *a\b* → <em>a\b</em>, *a\b\** → <em>a\b*</em>, _a\\_ → <em>a\</em>, and *a\* → no emphasis. 2.1–2.3 fail without the fix; all pass on 9.1 and 8.6.
  • Suggested action apply prototype (or fold it into Torsten's branch, after fixing that branch's non-CommonMark expectations)
  • Suggested ticket comment A one-line regexp change (accept any backslash pair inside the emphasis body) fixes the reported cases and the \* case. Prototype with tests available. Some expectations in the branch's tests (*a\b* → ab) differ from cmark.
⬜cd5c0748402025Bearer Authentication in create_interfacepatchREPRODUCED (9.1, 8.6): rest.man already documents auth bearer for create_interface, and rest::simple supports it, but create_interface ignores it…

Open ticket cd5c074840 · Patch · rest · filed 2025-12-10

Full verdict

REPRODUCED (9.1, 8.6): rest.man already documents auth bearer for create_interface, and rest::simple supports it, but create_interface ignores it. No Authorization header is sent.

  • Module rest 1.8
  • Tested local HTTP server recording request headers; interface with auth {bearer ...}.
  • Cause rest.tcl:create_interface handles only basic and sign.
  • Prototype prototypes/cd5c074840.patch (109 lines; adds a new modules/rest/rest.test, test rest-2.0). It uses lappend proc [list lappend headers Authorization "Bearer $token"]. The ticket's version puts the token into a double-quoted string inside the generated proc body, so a token containing $, [ or " is substituted or evaluated at call time. The test uses such a token. It fails without the fix and passes on 9.1 and 8.6. This patch and the 3613725fff patch both create rest.test (and the cdbdd399e4 patch too), so the tests need merging if more than one is applied.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: bearer is documented for create_interface but not implemented. Proposed fix adds it with list quoting, so tokens with Tcl special characters stay intact; it has a test.
🔧8ac37e504d2025No longer "-encoding binary" in Tcl 9.0patchREPRODUCED (9.0, 9.1; 8.6 is fine). distribution.snapshot ScmUnpack fails with unknown encoding "binary" for .tgz/.gz archives. Open ticket 146…

Open ticket 8ac37e504d · Bug · A Category Is Missing · filed 2025-10-07

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in c306c66ec1 on tcl9-encoding-binary, not merged: Fix 8ac37e504d, 146ae829a9: -encoding binary is gone in Tcl 9. practcl's snapshot unpacking and doctools::tcl::parse file still used it, so they failed on T
  • 🔧 2026-11-15 check-in 436138f6ad on tcl9-encoding-binary, not merged: Close the archive channel that ScmUnpack leaves open before deleting the test files, so that practcl-snapshot-unpack-001 also passes on Windows.
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch tcl9-encoding-binary: the snapshot unpacking in practcl uses -translation binary. The same branch fixes doctools::tcl::parse (146ae829a9). Tests practcl-snapshot-unpack-001 and doctools-tcl-parse-*-12.*.
Full verdict

REPRODUCED (9.0, 9.1; 8.6 is fine). distribution.snapshot ScmUnpack fails with unknown encoding "binary" for .tgz/.gz archives. Open ticket 146ae829a9 (doctools) reports the same leftover in doctools2base/tcl_parse.tcl:44: doctools::tcl::parse file always fails on Tcl 9. These were missed by the 5da15eba71 cleanup (2024).

  • Module practcl 0.16.6 (practcl.tcl:7676, and its source build/class/distro/snapshot.tcl:34)
  • Tested direct repro of the fconfigure call on 9.0/9.1; new tests with sak on 9.1 and 8.6. Both new tests fail on 9.1 without the fix and pass on 8.6. With the fix all pass. The doctools-msgcat-2.0 failure was already there before the patch.
  • Cause fconfigure $fh -encoding binary -translation lf in practcl, and fileutil::cat -translation binary -encoding binary in doctools2base
  • Prototype prototypes/8ac37e504d.patch (130 lines; fixes 3 code lines). It uses -translation binary in practcl (both copies) and drops the redundant -encoding binary in tcl_parse.tcl. Tests: practcl-snapshot-unpack-001 (practcl.test) and doctools-tcl-parse-*-12.0 (tcl_parse.test). Two leftovers are not touched: amazon-s3/S3.test:517,543 (constraint-gated tests) and the example in jpeg/jpeg.man:63.
  • Suggested action apply prototype, close this one and 146ae829a9 as Fixed
  • Suggested ticket comment Confirmed on Tcl 9. practcl now opens snapshot archives with -translation binary. The same fix covers doctools2base/tcl_parse.tcl from 146ae829a9.
🔧146ae829a92025No longer "-encoding binary" in Tcl 9.0patchREPRODUCED (9.1, 9.0; works on 8.6)

Open ticket 146ae829a9 · Bug · doctools · filed 2025-10-07

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in c306c66ec1 on tcl9-encoding-binary, not merged: Fix 8ac37e504d, 146ae829a9: -encoding binary is gone in Tcl 9. practcl's snapshot unpacking and doctools::tcl::parse file still used it, so they failed on T
  • 🔧 2026-11-15 check-in 436138f6ad on tcl9-encoding-binary, not merged: Close the archive channel that ScmUnpack leaves open before deleting the test files, so that practcl-snapshot-unpack-001 also passes on Windows.
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch tcl9-encoding-binary, together with 8ac37e504d: doctools::tcl::parse file no longer passes -encoding binary. The new tests parse the existing data files.
Full verdict

REPRODUCED (9.1, 9.0; works on 8.6)

  • Module doctools::tcl::parse 0.2 (doctools2base/tcl_parse.tcl)
  • Tested fileutil::cat -translation binary -encoding binary fails on 9.0/9.1 with unknown encoding "binary", so doctools::tcl::parse file always fails on Tcl 9. The test suite only exercised parse text.
  • Cause tcl_parse.tcl:44 typemethod file still passes -encoding binary. The 2024 sweep (ticket 5da15eba71, closed Fixed) missed this one.
  • Prototype prototypes/146ae829a9.patch (+16/-1, tests doctools-tcl-parse-*-13.N). It drops -encoding binary (-translation binary alone already selects iso8859-1, so the behaviour is the same on 8.6). The new tests run parse file over the existing tcl_data inputs. Without the fix all 18 fail on 9.1; with it they pass on 9.1 and 8.6. The only remaining failure, doctools-msgcat-2.0, also fails unpatched (see 1968777fff).
  • Related 8ac37e504d (same report for practcl.tcl:7676, still has -encoding binary, other batch). 0c118eed38 (tclvfs, closed Invalid). 0e69b86c2e (Tkhtml, not tcllib).
  • Suggested action apply the prototype
  • Suggested ticket comment Confirmed: doctools::tcl::parse file fails on Tcl 9 because of -encoding binary in tcl_parse.tcl. Fix: drop it, since -translation binary already implies iso8859-1. A test is added.
🔧1a7976b8732025pluginmgr potential collisions with files/folders in cwdpatchREPRODUCED (9.1, 8.6) for the directory case. The file case is documented behaviour: "if the string contains the path to an existing file then this…

Open ticket 1a7976b873 · Bug · (unused) · filed 2025-10-06

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 21e2f1767a on pluginmgr-cwd-dir-collision, not merged: Fix 1a7976b873: pluginmgr failed to load a plugin when the current directory had a directory with the plugin's name. Use file isfile instead of file exists. A
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch pluginmgr-cwd-dir-collision: LoadPlugin uses file isfile instead of file exists, so a directory with the plugin's name in the current directory no longer breaks loading. Adds pluginmgr.test.
Full verdict

REPRODUCED (9.1, 8.6) for the directory case. The file case is documented behaviour: "if the string contains the path to an existing file then this file is taken as the implementation". The reporter says part of the report was pilot error.

  • Module pluginmgr 0.5
  • Tested new test: a plugin package on auto_path plus a directory with the plugin's name in cwd. load simple fails with couldn't read file "simple": is a directory.
  • Cause pluginmgr.tcl:LoadPlugin uses file exists $name, which is also true for directories
  • Prototype prototypes/1a7976b873.patch (94-line patch, +53/-2: file exists -> file isfile, .man wording, new pluginmgr.test with test pluginmgr-load-1.0, which fails without the fix). The patch adds the test file; the module had no tests.
  • Suggested action apply prototype. Restricting direct loads to the configured paths would change documented API: maintainers' decision (see also dffc63972b).
  • Suggested ticket comment A directory in the cwd with the plugin's name breaks load, because LoadPlugin checks file exists. A proposed fix uses file isfile, with a test. Loading a file path relative to the cwd is documented behaviour.
⬜8ab3be2e6a2025list command overwrites $plugin variablepatchREPRODUCED (9.1, 8.6)

Open ticket 8ab3be2e6a · Bug · pluginmgr · filed 2025-10-05

Full verdict

REPRODUCED (9.1, 8.6)

  • Module pluginmgr 0.5
  • Tested pm load x; pm list; pm plugin → y (the last listed plugin) on 9.1 and 8.6, using Tcl-module plugins. Side finding: package (pkgIndex) based plugins do not load at all on 9.x. TIP 579 made the safe base stop syncing the child's auto_path with interpAddToAccessPath, which SetupIp/path rely on. It works on 8.6. No ticket exists for this.
  • Cause pluginmgr.tcl method list uses the instance variable plugin as the regexp capture variable.
  • Prototype prototypes/8ab3be2e6a.patch (43 changed lines, mostly the new pluginmgr.test, which did not exist before; test pluginmgr-ticket-8ab3be2e6a). The capture variable is renamed. The test fails without the fix on 9.1 and 8.6.
  • Suggested action apply prototype. Consider a new ticket for the Tcl 9 auto_path regression (fix: append the access-path tokens to the child's auto_path, or use ::safe::interpConfigure -autoPath).
  • Suggested ticket comment Confirmed; the regexp in list captured into the instance variable plugin. A proposed fix, with a new pluginmgr.test, is ready.
⬜dffc63972b2025pluginmgr doesn't find modules in subdirectoriespatchREPRODUCED (9.1, 8.6)

Open ticket dffc63972b · Bug · pluginmgr · filed 2025-10-05

Full verdict

REPRODUCED (9.1, 8.6)

  • Module pluginmgr 0.5
  • Tested Tcl modules pmsub/plugins/{p,q}-1.0.tm with -pattern pmsub::plugins::*: load p works but list returns nothing on 9.1 and 8.6. The "proj* gives ect" remark is expected behaviour: list returns the part matched by *, which load maps back.
  • Cause pluginmgr.tcl method list triggers the package scan with package require bogus-package. The tm unknown handler then only scans the top level of the module roots, never pmsub/plugins/.
  • Prototype prototypes/dffc63972b.patch (62 changed lines, test pluginmgr-ticket-dffc63972b). It derives the bogus name from the pattern, as the reporter suggested (string map {* bogus-package} $pattern). The test fails without the fix on 9.1 and 8.6. This patch is stacked on 8ab3be2e6a.patch (it includes it, because both use the new pluginmgr.test).
  • Suggested action apply prototype (after 8ab3be2e6a)
  • Suggested ticket comment Confirmed for Tcl modules in nested namespaces. A proposed fix that derives the bogus package name from -pattern, as suggested, is ready with a test.
🔧7bd2d439ca2025"crc32 -filename $name" end in error on files with special content (not encoding problem)patchREPRODUCED (emulated, critcl path only); the pure Tcl path works on Windows (9.1, 8.6): crc::crc32 -filename of a 4800-byte file equals the strin…

Open ticket 7bd2d439ca · Bug · crc · filed 2025-08-04

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 1fb1b957bd on crc32-critcl-negative-sum, not merged: Fix 7bd2d439ca: with tcllibc on Tcl 9, crc32 of data over one chunk (4 KiB for files) failed where C long is 32 bits (Windows): Crc32_c returns a negative val
  • 💬 2026-11-15 serhiy.storchaka: This comes from the critcl implementation in tcllibc, where C long is 32 bits (Windows). Crc32_c returns a negative value, and Crc32Update stores it unmasked. On the next chunk (4 KiB for files) Tcl 9 rejects it as the seed. Tcl 8.6 accepted it, and the pure Tcl implementation works, which explains the different results above. Proposed fix in branch crc32-critcl-negative-sum: mask the running sum, and close the fi…
Full verdict

REPRODUCED (emulated, critcl path only); the pure Tcl path works on Windows (9.1, 8.6): crc::crc32 -filename of a 4800-byte file equals the string CRC (D1A741BC). The critcl path (tcllibc), which the triage found to be the cause, could not be tested because tcllibc is not built on the VM.

  • Module crc32 1.3.4 + critcl crcc (tcllibc)
  • Tested test crc32-7.0 replaces Crc32_c with a mock that behaves like the C code with a 32-bit long under Tcl 9's Tcl_GetLongFromObj range (LONG_MIN..ULONG_MAX). It fails without the fix and passes with it on 9.1 and 8.6.
  • Cause crc32c.tcl Crc32_c returns Tcl_NewLongObj(t ^ 0xFFFFFFFF), which is negative where long has 32 bits. crc32.tcl Crc32Update then stores $sum ^ 0xFFFFFFFF without masking, which gives values like -3632003085. On the next chunk Tcl 9 rejects such a seed with "integer value too large to represent". Tcl 8.6 accepted -ULONG_MAX..ULONG_MAX, which is why 8.6 worked. Harald probably had no tcllibc loaded, so he got the pure Tcl path. In crc32 -filename, an error also left the file open and the token array behind.
  • Prototype prototypes/7bd2d439ca.patch (83 lines, tests crc32-7.0, crc32-7.1). Masks the sum to 32 bits in Crc32Update, and closes the file and drops the token when an error happens. A cleaner long-term fix would use Tcl_WideInt in crc32c.tcl too.
  • Suggested action apply prototype; ask the reporter to confirm on Windows
  • Suggested ticket comment Cause found: on Windows (32-bit long) the critcl Crc32_c returns negative values, and Tcl 9's stricter Tcl_GetLongFromObj then rejects the unmasked sum for the second 4 KiB chunk. Tcl 8.6 accepted it. Proposed fix masks the running sum and closes the file on error.
Windows VM, run 1
  • Verdict REPRODUCED (emulated, critcl path only); the pure Tcl path works on Windows (9.1, 8.6): crc::crc32 -filename of a 4800-byte file equals the string CRC (D1A741BC). The critcl path (tcllibc), which the triage found to be the cause, could not be tested because tcllibc is not built on the VM.
  • Tested tclsh91 and tclsh86t with tcllib/modules on auto_path.
🔧34a589f6602025could not load ftpd in Tcl 9.0patchREPRODUCED (9.1, 9.0; 8.6 loads fine)

Open ticket 34a589f660 · Bug · ftpd · filed 2025-07-22

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in db1cb608a2 on ftpd-tcl9-load, not merged: Fix 34a589f660: package require ftpd failed on Tcl 9 with can't read "tcl_platform(user)". The namespace body used 'global tcl_platform', which does nothing o
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch ftpd-tcl9-load: the change suggested in the ticket, using $::tcl_platform(user) instead of global tcl_platform, which does nothing in a namespace body. Adds ftpd.test.
Full verdict

REPRODUCED (9.1, 9.0; 8.6 loads fine)

  • Module ftpd 1.4.1
  • Tested package require ftpd on 9.1 and 9.0 fails with can't read "tcl_platform(user)".
  • Cause ftpd.tcl, top-level namespace eval ::ftpd uses global tcl_platform, which does nothing outside a proc. Tcl 9 no longer falls back to global variables in namespace context.
  • Prototype prototypes/34a589f660.patch (1 line, the reporter's fix; new modules/ftpd/ftpd.test with test ftpd-1.0)
  • Suggested action apply prototype, close as Fixed (Harald acknowledged it in 2025-07 but no check-in yet)
  • Suggested ticket comment Proposed fix (the one from the report) with a test in the prototype; ftpd now gets a test file.
⬜d3f66f9a752025mime::field_decode remove spaces in specific situationspatchREPRODUCED (9.1, 8.6)

Open ticket d3f66f9a75 · Bug · mime · filed 2025-05-19

Full verdict

REPRODUCED (9.1, 8.6)

  • Module mime 1.7.2
  • Tested reporter's Subject → "...GeneralsekretärKlüssendorf ... vorKonflikt"; field_decode "=?UTF-8?Q?a_?= =?UTF-8?Q?b?=" → "ab"
  • Cause mime.tcl:qp_decode maps "_" to space before trimming trailing whitespace, so an encoded trailing space in a Q encoded word is dropped
  • Prototype prototypes/d3f66f9a75.patch (68 lines; the reporter's reordering; tests mime-18.1 (reporter's subject), mime-18.2 (word_decode/field_decode with trailing "_"); both fail without the fix; mime 102→104 pass on 9.1 and 8.6)
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: qp_decode turned "_" into a space and then trimmed it. A proposed fix applies the trimming first, with a test.
⬜a0cbdc7e6f2025zipfile::decode::unzip permissions bug with some zipspatchREPRODUCED (9.1, 8.6)

Open ticket a0cbdc7e6f · Bug · zip · filed 2025-03-06

Full verdict

REPRODUCED (9.1, 8.6)

  • Module zipfile::decode 0.11.0
  • Tested zip with MS-DOS external attributes 0x20 (archive) and 0x21 (archive+readonly), MSW = 0. Extracted files get mode 0000 (unreadable) on 9.1 and 8.6. Unix-made zips still need their permissions applied, and the new test 8.1 checks that.
  • Cause decode.tcl CopyFile applies (efattr >> 16) & 0x1ff as permissions whenever efattr != 0. In DOS/Windows zips only the low byte is used. The condition should test the MSW. Side finding: VER uses >> 16 instead of >> 8, so the "version made by" host is always reported as MS-DOS. That would need fixing before the host byte can be used instead.
  • Prototype prototypes/a0cbdc7e6f.patch (61 changed lines, mostly the shared mkTestZip helper; tests zipfile-decode-unzip-8.0, 8.1 in encode.test, constraint unix). Permissions are applied only when efattr >> 16 is non-zero. 8.0 fails without the fix. The full zip suite passes on 9.1 and 8.6.
  • Suggested action apply prototype (merge the helper with the c976e4a85d patch). Consider fixing VER separately.
  • Suggested ticket comment Unix permissions were applied from the high 16 bits even for MS-DOS/Windows attributes, which gave mode 000. Proposed fix: permissions are applied only when the high word is set. Tests 8.0/8.1 in encode.test.
⬜e70a9c8db92025json.tab.c: redefinition of typedef 'YYSTYPE'patchREPRODUCED (gcc -std=c99 -pedantic warns "redefinition of typedef 'YYSTYPE'". clang -std=c99 warns that this is a C11 feature [-Wtypedef-redefi…

Open ticket e70a9c8db9 · Bug · json · filed 2025-02-16

Full verdict

REPRODUCED (gcc -std=c99 -pedantic warns "redefinition of typedef 'YYSTYPE'". clang -std=c99 warns that this is a C11 feature [-Wtypedef-redefinition]. Older clang makes it an error.)

  • Module jsonc 1.1.2 (modules/json/c/json.y, json.tab.c)
  • Tested syntax-only compile of json.tab.c with gcc and clang against 9.1 headers, C99 and C11
  • Cause json.y user prologue has typedef union YYSTYPE YYSTYPE; and bison emits the same typedef again
  • Prototype prototypes/e70a9c8db9.patch (8 lines). Forward-declares union YYSTYPE; and uses union YYSTYPE * in the jsonlexp prototype, in both json.y and the generated json.tab.c. Clean under -std=c99 -pedantic with gcc and clang. No regression test possible (compile-time only; critcl build is not covered by the Tcl testsuite).
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed with -std=c99. A fix (forward declaration instead of a second typedef) is ready.
⬜57ee1f3e1d2024yaml2dict loses contents when values contain bracespatchREPRODUCED (9.1, 8.6)

Open ticket 57ee1f3e1d · Bug · yaml · filed 2024-11-19

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested dict2yaml {k {{val}}} emits k: {val}, which is a flow mapping in YAML, so yaml2dict returns k {}. Quoted "{val}" parses fine.
  • Cause yaml.tcl ::yaml::_dumpScalar emits plain scalars even when they start with an indicator character ({ [ & * ! # " ' % @ , | >`).
  • Prototype prototypes/57ee1f3e1d.patch (49 lines, tests yaml-36.1, yaml-36.2). Double-quotes single-line values that start with an indicator, escaping \ and ". Side finding: huddle create k {[a "b"]} itself errors (huddle::isHuddle runs lindex on non-list strings); not fixed here.
  • Suggested action apply prototype; maybe open a separate huddle ticket for the isHuddle problem
  • Suggested ticket comment The dumper wrote {val} unquoted, which YAML reads as a mapping. Proposed fix quotes values that start with an indicator; tests yaml-36.*.
⬜c976e4a85d2024zipfile::decode::unzip bug with some zips in ::close [::open $dst w] ; # touchpatchREPRODUCED (9.1, 8.6)

Open ticket c976e4a85d · Bug · zip · filed 2024-10-16

Full verdict

REPRODUCED (9.1, 8.6)

  • Module zipfile::decode 0.11.0
  • Tested zips generated with Python that have a directory entry d/sub without a trailing slash, marked as a directory either by the MS-DOS attribute (0x10) or by the unix mode S_IFDIR, plus a file d/sub/f.txt. unzipfile fails with couldn't open "out/d/sub": is a directory on 9.1 and 8.6. The reporter's archive (from GiD on Windows) probably has such entries, or entries ending in a backslash.
  • Cause decode.tcl CopyFile recognizes directories only by a trailing /. Any other zero-size entry is "touched" as a file, and that fails when the directory already exists because a child was extracted first (array names order).
  • Prototype prototypes/c976e4a85d.patch (59 changed lines, mostly the test helper mkTestZip that writes a stored zip with chosen vmb/efattr; tests zipfile-decode-unzip-7.0, 7.1 in encode.test). An entry is also treated as a directory when its name ends in / or \, or efattr has 0x10, or the unix file type is S_IFDIR. Both tests fail without the fix. The full zip suite passes on 9.1 and 8.6.
  • Suggested action apply prototype. Note: the a0cbdc7e6f patch adds the same helper to encode.test, so the second one to be applied needs a trivial merge.
  • Suggested ticket comment Directory entries without a trailing "/" (marked as directories only by their attributes) were extracted as empty files. Proposed fix: the directory attribute bits are now checked too. Test 7.0/7.1 in encode.test.
⬜d638a04e5c2024[coroutine::util gets_safety] bugspatchREPRODUCED (9.1, 8.6)

Open ticket d638a04e5c · Bug · coroutine · filed 2024-09-12

Full verdict

REPRODUCED (9.1, 8.6)

  • Module coroutine 1.4
  • Tested new coroutine.test with chan pipe on 9.1 and 8.6: unpatched, a complete 20000-char line is returned with limit 10; an incomplete over-limit line hangs until the timeout; coroutine::util gets_safety is "unknown or ambiguous subcommand"; no errorcode. Patched: 9/9 pass on both
  • Cause coroutine.tcl ::coroutine::util::gets_safety checks chan pending input before reading (always 0 at first, and counts data after the newline), never checks the line length, and checks pending only after being woken; gets_safety is not in the ensemble export list
  • Prototype prototypes/d638a04e5c.patch (267 lines incl. a new coroutine.test of ~140 lines; module had no tests; tests coroutine-gets_safety-1.0..1.8): read first, check line length, check pending only for an incomplete line before yielding; errorcodes {COROUTINE GETS_SAFETY LIMIT|TIMEOUT}; cancel the timer in finally; export gets_safety; fix doc typos and document timeout/errorcodes. Memory is still not strictly bounded for data that is already available (e.g. files), which is noted in the man page
  • Suggested action apply prototype (bump coroutine version on merge)
  • Suggested ticket comment gets_safety checked the pending input before reading and never checked the line length, and was missing from the ensemble. Proposed fix in a branch with new tests coroutine-gets_safety-*; it also adds error codes.
⬜91c114ee922024::math::linearalgebra::solveGauss gives overflow for matrices with a row of zerospatchREPRODUCED (9.1, 8.6). Worse than reported: even the non-singular solveGauss {{0 1} {1 0}} {1 2} gives "domain error", because solveGauss never p…

Open ticket 91c114ee92 · Bug · math :: linearalgebra · filed 2024-08-29

Full verdict

REPRODUCED (9.1, 8.6). Worse than reported: even the non-singular solveGauss {{0 1} {1 0}} {1 2} gives "domain error", because solveGauss never pivots.

  • Module math, math::linearalgebra 1.1.7
  • Tested the ticket example, {{1 2} {2 4}} and {{0 1} {1 0}} with solveGauss, and the ticket example with solvePGauss. All give "domain error: argument not in valid range" (9.1, 8.6).
  • Cause linalg.tcl. solveGauss divides by a zero pivot (no pivoting at all), and solveTriangular divides by a zero diagonal. Both produce Inf/NaN and then a domain error.
  • Prototype prototypes/91c114ee92.patch (132 lines, ~40 changed, tests solve-2.1..2.5). solveGauss swaps in a lower row when the pivot is exactly zero and errors "Matrix is singular" if none exists. solveTriangular errors "Matrix is singular" on a zero diagonal, which also covers solvePGauss. linalg.man is updated. The tests fail without the fix; linalg.test passes 131/131 and statistics.test (mvlinreg users) 154/154 on 9.1 and 8.6.
  • Suggested action apply prototype
  • Suggested ticket comment solveGauss fails even on non-singular matrices with a zero pivot, and singular matrices give an obscure domain error. A prototype adds a minimal row swap and a clear "Matrix is singular" error.
⬜2bcc38cc1d2024huddle2yaml of yaml package doesn't preserve string indentationpatchREPRODUCED (9.1, 8.6)

Open ticket 2bcc38cc1d · Bug · yaml · filed 2024-08-29

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested dict2yaml/huddle2yaml of indented code loses all indentation in the | block.
  • Cause yaml.tcl ::yaml::_doLiteralBlock applies string trim to every line.
  • Prototype prototypes/2bcc38cc1d.patch (51 lines, tests yaml-35.1, yaml-35.2). Uses string trimright; output parses back unchanged in tcllib and PyYAML. Limitation: a value whose *first* line is indented would still need a block indentation indicator; not handled.
  • Suggested action apply prototype
  • Suggested ticket comment Literal blocks stripped leading whitespace from each line. Proposed fix keeps relative indentation; tests yaml-35.*.
⬜d470bac96c2024httpd timeout handling (possibly related to SSL)patchREPRODUCED (9.1, 8.6)

Open ticket d470bac96c · Bug · httpd · filed 2024-06-03

Full verdict

REPRODUCED (9.1, 8.6)

  • Module httpd 4.3.6 (coroutine 1.x for gets_safety)
  • Tested a client sends an incomplete request line, then the pending ::httpd::coro::<uuid> timeout after-script is fired by hand. Result: can't read "query": no such variable, exactly as in the report. The SSL part is not a bug: the man page server is plain HTTP, so curl https:// hangs in the TLS handshake until the 120 s read timeout fires.
  • Cause httpd.tcl / build/server.tcl ::httpd::server method Connect: query is only created after the request line is read. The on error handler uses [dict getnull $query REQUEST_URI], so a read timeout (or any gets error) raises a background error instead of answering 400. The handler also used the wrong key (it should be http REQUEST_URI).
  • Prototype prototypes/d470bac96c.patch (101 lines, test httpd-client-0009). It initialises set query [dict create UUID $uuid] before the try and fixes the debug key. The test fails without the fix. Module tests pass on 9.1 and 8.6.
  • Suggested action apply prototype. Side note: coroutine::util::gets_safety rethrows read errors with return -code $result -options $opts, which loses the error message (separate small bug in modules/coroutine).
  • Suggested ticket comment The error comes from the Connect error handler reading query before it exists, after the 120 s read timeout. Your curl used https against a plain-HTTP server, hence the stall. Proposed fix initialises query first (patch + test httpd-client-0009).
⬜00c6d3f8522024yaml.tcl typo in huddle superclass keywordpatchREPRODUCED (9.1, 8.6)

Open ticket 00c6d3f852 · Bug · yaml · filed 2024-03-29

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested huddle equal on two yaml2huddle results fails with Invalid subcommand 'equal' for type '::huddle::Type_str'.
  • Cause yaml.tcl ::yaml::_makeChildType writes superClass string; huddle looks for superclass.
  • Prototype prototypes/00c6d3f852.patch (39 lines, test yaml-33.1)
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed; with superclass the yaml scalar types inherit huddle string methods (e.g. huddle equal). Proposed fix with test yaml-33.1.
⬜6abf74eec82024md5crypt failurepatchREPRODUCED (9.1, 9.0: error; 8.6: wrong hash). The test md5crypt-5.0/5.1 is now marked knownBug, so it is skipped in normal runs.

Open ticket 6abf74eec8 · Bug · md5crypt · filed 2024-03-25

Full verdict

REPRODUCED (9.1, 9.0: error; 8.6: wrong hash). The test md5crypt-5.0/5.1 is now marked knownBug, so it is skipped in normal runs.

  • Module md5crypt 1.1.0 (md5crypt.tcl, pure Tcl)
  • Tested 9.1: md5crypt::md5crypt "пароль" abcdefgh → "expected code point values below 0xff"; the test's salt "￾\0\1\2" gives the same error. 8.6: returns $1$abcdefgh$Oi/N6O.fyd5T7zv71xnU10, but openssl passwd -1 (UTF-8) gives $1$abcdefgh$QfFdDZYjOC5k44ZZ.ueGu0. A Latin-1 "pässwörd" also differs from openssl on both versions.
  • Cause md5crypt.tcl:md5crypt_tcl feeds the password and salt strings straight into md5 v2 (binary), so chars >U+00FF fail on Tcl 9 and are truncated on 8.6. The critcl version (md5cryptc, fixed in 00c0fc2344) uses the UTF-8 string rep.
  • Prototype prototypes/6abf74eec8.patch (65 lines; encode pw and salt with encoding convertto utf-8 in md5crypt_tcl; tests md5passwd-8.1–8.3 (Ukrainian and Latin-1 passwords, md5crypt and aprcrypt, expected values from openssl), md5passwd-9.* for the pure-Tcl path when C is loaded; 3 fail without fix on 9.1 and 8.6, all pass with it)
  • Suggested action apply prototype. Note that Latin-1 passwords now hash as UTF-8 on 8.6 too, which is a compatibility change; it matches openssl/glibc and the C accelerator. md5crypt-5.0/5.1 stay knownBug: with the fix they no longer error, but the expected hash in the test came from the C code's modified UTF-8 (\0 → C0 80) and is questionable.
  • Suggested ticket comment The pure-Tcl md5crypt hashed the raw characters, so non-Latin-1 passwords errored on Tcl 9 and gave non-standard hashes on 8.6. A proposed fix hashes UTF-8 (like the C version and openssl), with tests.
⬜4f25e933fc2024yaml / json2huddlepatchREPRODUCED (9.1, 8.6)

Open ticket 4f25e933fc · Bug · yaml · filed 2024-03-25

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2 / huddle 0.5 / huddle::json 0.2
  • Tested after package require yaml, huddle null fails with Invalid subcommand 'null' for type '::huddle::Type_null', and so does huddle::json2huddle {{"a": null}}. Without yaml both work.
  • Cause yaml.tcl registers its !!null type from namespace ::yaml::types::null. huddle addType derives the ensemble name from the namespace tail, so it overwrites huddle's own ::huddle::Type_null. (The reporter's patches are phantom artifacts in the repo, so I couldn't read them.)
  • Prototype prototypes/4f25e933fc.patch (44 lines, tests yaml-34.1, yaml-34.2). Registers the yaml type as yaml_null (tag still !!null). Side effect: huddle type of a parsed YAML null becomes yaml_null instead of null. Needs a human decision on the name.
  • Suggested action apply prototype after deciding on the type name; it goes with 00c6d3f852 (the reporter's second finding)
  • Suggested ticket comment yaml's null type replaced huddle's own null type. Proposed fix registers it under a distinct name; tests yaml-34.*.
⬜c0e70192dd2023httpd ::httpd::content.file method content issuepatchREPRODUCED (9.1, 8.6)

Open ticket c0e70192dd · Bug · httpd · filed 2023-11-10

Full verdict

REPRODUCED (9.1, 8.6)

  • Module httpd 4.3.6
  • Tested content.file with path set to a directory containing sub/index.html (and sub/index.md). GET /docs/sub/index.html returns "Listing contents of /sub/index.html/" instead of the file (standalone script and module test, 9.1 and 8.6).
  • Cause httpd.tcl / build/file.tcl content.file method content: the test [file isdirectory $local_file] || [file tail $local_file] in {index index.html index.tml index.md} treats any existing index file as a directory. It then searches for index files inside the file and falls back to DirectoryListing. FileName already maps a top-level "index.html" to the directory, so the tail test is never needed. The fix hypnotoad suggested in the ticket is wrong, because it would treat every non-index file as a directory.
  • Prototype prototypes/c0e70192dd.patch (68 lines, test httpd-client-0010). The condition is reduced to [file isdirectory $local_file]. The test uses index.md because in the test harness ::fileutil::magic::filetype is missing, so .html files return a 500 error there (side issue). The test fails without the fix (0 1). Module tests pass on 9.1 and 8.6.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: an index file in a subdirectory is treated as a directory and you get a listing. Proposed fix: decide on file isdirectory only (patch + test httpd-client-0010). Note that the workaround suggested in the 2024-01-09 comment inverts the logic for all other files.
⬜a316389e992023math::decimal doesn't handle Tip 551 underscores in numbers fullypatchREPRODUCED (9.1, 8.6)

Open ticket a316389e99 · Bug · math · filed 2023-08-25

Full verdict

REPRODUCED (9.1, 8.6)

  • Module math, math::decimal 1.0.5
  • Tested compare [fromstr 1.000001] [fromstr 1.000_001] gives 1 (9.1, 8.6). fromstr accepts any garbage silently: abc → {0 abc 0}, 0x10, 12a, 1.2.3 → 123e-3, 1e1_0 → exponent 1. On 8.6 fromstr 09 then fails in arithmetic with an octal error.
  • Cause decimal.tcl:fromstr parses by string slicing without any validation, so _ stays inside the mantissa and is counted as a fraction digit.
  • Prototype prototypes/a316389e99.patch (95 lines, tests decimal-fromstr-2.1, 2.2, 2.3.1–2.3.8). It validates the input with a regexp (sign, digits with optional _ between digits, fraction, exponent, Inf/NaN), strips the underscores, strips leading zeros of integers (the octal issue), and errors on anything else. decimal.man documents the accepted syntax. 9–10 tests fail without the fix; all pass on 9.1 and 8.6.
  • Suggested action human decision. Erroring on invalid input is a behaviour change (aku asked for that decision); if accepted, apply prototype.
  • Suggested ticket comment fromstr does no validation, so underscores (and any garbage) end up in the mantissa. A prototype accepts TIP 551 underscores and rejects non-decimal input with an error.
⬜66063baec42023ncgi cannot handle 4-byte utf-8 characterspatchREPRODUCED (9.1, 8.6). %F0%A0%AE%B7 decodes to 4 Latin-1 characters (re-encoded as c3b0c2a0c2aec2b7) instead of U+20BB7.

Open ticket 66063baec4 · Patch · ncgi · filed 2023-02-10

Full verdict

REPRODUCED (9.1, 8.6). %F0%A0%AE%B7 decodes to 4 Latin-1 characters (re-encoded as c3b0c2a0c2aec2b7) instead of U+20BB7.

  • Module ncgi 1.4.6
  • Tested the ticket's script on 9.1 and 8.6. On 8.6 the patched result is a surrogate pair that converts back to f0a0aeb7.
  • Cause ncgi.tcl:decode has regsubs for 3- and 2-byte UTF-8 sequences only
  • Prototype prototypes/66063baec4.patch (42 lines, 2 code lines plus tests ncgi-3.11/3.12, which compare UTF-8 bytes so they also work on 8.6. Both fail without the fix.) The attached diff.patch is not in the local repo (phantom artifact), so it could not be compared.
  • Suggested action apply prototype
  • Suggested ticket comment decode now also handles 4-byte UTF-8 sequences (F0–F4 lead bytes).
🔧a1dec8173c2022des package missing procpatchREPRODUCED (9.1, 8.6): DES::des -key … -in $chan gives invalid command name "Chunk".

Open ticket a1dec8173c · Bug · des · filed 2022-08-21

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in eff01ad452 on des-chunk-proc, not merged: Fix a1dec8173c: DES::Update and the -chunksize option called the missing proc ::DES::Chunk. Add it, in the style of blowfish: process whole blocks, keep the r
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch des-chunk-proc: adds DES::Chunk, in the style of blowfish, so -in and -chunksize work. Tests des-10.1 and des-10.2.
Full verdict

REPRODUCED (9.1, 8.6): DES::des -key … -in $chan gives invalid command name "Chunk".

  • Module des 1.2
  • Tested tclsh9.1, tclsh8.6; des.test
  • Cause des.tcl ::DES::des installs a fileevent handler Chunk that was never written (blowfish and aes have one).
  • Prototype prototypes/a1dec8173c.patch (87 lines, tests des-10.1, des-10.2). Adds ::DES::Chunk in the style of blowfish: it processes whole blocks, keeps the remainder, zero-pads only the final partial block (same as data mode), and reports errors. It is a bit over the 60-line guideline, mostly because of the tests.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: -in never worked because Chunk was missing. Proposed patch adds it (same approach as blowfish), with tests for -in and -out.
🔧1dd19ccd142022SMTPd Delivery Callback ErrorpatchREPRODUCED (9.1, 8.6)

Open ticket 1dd19ccd14 · Patch · smtpd · filed 2022-06-30

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in e3ea101883 on smtpd-deliver-command-prefix, not merged: Fix 1dd19ccd14: the smtpd -deliver callback worked only with a single-word command. Call it as a command prefix with {*}. Add smtpd.test.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch smtpd-deliver-command-prefix: -deliver is called as a command prefix with {*}. The eval suggested in the ticket would break on message data with spaces. Adds smtpd.test.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module smtpd 1.6
  • Tested in-process smtpd with -deliver [list deliverto delivered]. Delivery fails with 554 (invalid command name "deliverto delivered").
  • Cause smtpd.tcl: deliver_old calls $deliver ... as a single word, unlike the other callbacks. The ticket's proposed eval $deliver [state ...] would break on data with spaces.
  • Prototype prototypes/1dd19ccd14.patch (117 lines, mostly the new file modules/smtpd/smtpd.test; test smtpd-1.1). Uses {*}$deliver. Passes on 9.1 and 8.6, fails without the fix.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: -deliver only works with a single-word command. Fixed with {*} (the suggested eval would break on message data with spaces). Patch with test in prototypes/1dd19ccd14.patch.
⬜9a503797fa2022yaml::yaml2dict doesn't support keys with literal colons in thempatchREPRODUCED (9.1, 8.6)

Open ticket 9a503797fa · Bug · yaml · filed 2022-05-26

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested the ticket example drops message:PHYS-279; "a:b: c" gives the scalar a:b: c instead of {a:b c}; "http://x: 1" is not a mapping.
  • Cause yaml.tcl _parsePlainScalarInBlock and _parsePlainScalarInFlow: the plain-scalar regexps let :[^ ]+ swallow the key's final :, so the key indicator is lost.
  • Prototype prototypes/9a503797fa.patch (55 lines, tests yaml-38.1, yaml-38.2). It changes both regexps so a colon inside a scalar must be followed by a non-colon character. Results match PyYAML (a::b, http://x.org/a, e:f still work). It touches the same line as a23348be0b.
  • Suggested action apply prototype
  • Suggested ticket comment The plain-scalar pattern consumed the key's final colon. Proposed fix with tests yaml-38.*.
🔧e2ba08b2852022huddle compile generates incorrect huddle for boolean typepatchREPRODUCED (9.1, 8.6)

Open ticket e2ba08b285 · Bug · huddle · filed 2022-05-17

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 33e681a995 on huddle-compile-bool-tag, not merged: Fix e2ba08b285: huddle compile bool used the tag "bool" instead of "b".
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch huddle-compile-bool-tag: compile uses the boolean type's tag "b" instead of "bool". Test huddle-3.10.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml/huddle 0.5
  • Tested huddle compile bool true gives HUDDLE {bool true}, and jsondump then fails with "is not a huddle". The same happens nested (compile {dict a bool} ...). The bug is present since 1.18.
  • Cause huddle.tcl ::huddle::compile, branch bool: it wraps with tag bool, but the boolean type's tag is b.
  • Prototype prototypes/e2ba08b285.patch (44 lines, test huddle-3.10). The tag is now b. The test fails without the fix. Yaml module tests pass on 9.1 and 8.6. Test-isolation side note: huddle-12.0 deletes ::map but leaves the type registered, so any jsondump in a later test fails with invalid command name "::huddle::Type_map". New tests must therefore go before huddle-12.0.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: compile uses tag "bool" instead of the boolean type's tag "b". One-line fix with test huddle-3.10.
⬜57f151c3542022HTML block elements mistreated when used in inline markdownpatchREPRODUCED (9.1, 8.6)

Open ticket 57f151c354 · Bug · markdown · filed 2022-05-05

Full verdict

REPRODUCED (9.1, 8.6)

  • Module markdown, Markdown 1.2.5
  • Tested the ticket input (tests div-1.3/1.4 are on trunk under the knownbug constraint since bf7b7ef0cc). <div allowfullscreen></div> on a paragraph line becomes &lt;div allowfullscreen&gt;</div> inside <p>. With ="1" the tag is kept but stays inside the paragraph.
  • Cause markdown.tcl:parse_block. The paragraph loop does not stop at a line that starts a block-level HTML tag. Also, the inline re_htmltag in parse_inline does not accept attributes without a value (the block regexp was fixed for that in 31b56b6b47; the inline one was not).
  • Prototype prototypes/57f151c354.patch (134 lines, ~25 changed). A line starting with <div|p|h1-6|blockquote|pre|table|dl|ol|ul|form|fieldset|iframe|math now ends the paragraph, with a blank-line separator as the tests expect. The inline tag regexp now accepts valueless attributes. The patch removes the knownbug constraint from div-1.3/1.4 and adds div-1.5 (<span hidden> inline) and div-1.6 (an inline tag at line start does not break the paragraph). div-1.3/1.4/1.5 fail without the fix; on 9.1 and 8.6 the result is 89 passed, 6 skipped.
  • Suggested action apply prototype
  • Suggested ticket comment The paragraph parser does not stop at block-level HTML tags, and the inline tag regexp rejects valueless attributes. Proposed fix in prototype; it enables the existing div-1.3/1.4 known-bug tests.
⬜0d824f26fb2022Markdown mishandles escaped characters in some casespatchREPRODUCED (9.1, 8.6)

Open ticket 0d824f26fb · Bug · markdown · filed 2022-03-30

Full verdict

REPRODUCED (9.1, 8.6)

  • Module markdown, Markdown 1.2.5
  • Tested Markdown::convert on cases 9, 11, 16, 17 from the ticket (9.1, 8.6). They are still marked .FAIL and skipped in markdown.test (since 4a1e5f6187, 2022-05-03). x __\.__ x gives x <em></em>.<em></em> x, and [foo_bar](sna\_fu) gives href="sna\_fu".
  • Cause markdown.tcl:parse_inline. (a) The emphasis regexp only lets a backslash appear in the body when it escapes the delimiter itself (\\$chr), so __\.__ does not match as strong emphasis. (b) Inline-link URLs are only html-escaped; backslash escapes are not removed.
  • Prototype prototypes/0d824f26fb.patch (78 lines; the four .FAIL cases basic-1.9/11/16/17 are re-enabled). It allows \\. (any escaped char) in the emphasis body and adds a small unescape helper for link destinations. The emphasis part is the same change as in a028a78b5b.patch. Without the fix the four tests fail; with it markdown.test passes on 9.1 and 8.6 (89 passed, 8 skipped).
  • Suggested action apply prototype (together with a028a78b5b, or instead of it; they overlap in one line)
  • Suggested ticket comment Emphasis regexp rejects backslashes that do not escape the delimiter, and link destinations are not unescaped. Proposed fix in prototype that enables the four known-bug tests.
⬜800c0dc7c12022yaml::yaml2dict inserts spaces on quoted line continuationpatchREPRODUCED (9.1, 8.6)

Open ticket 800c0dc7c1 · Bug · yaml · filed 2022-03-30

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested "alphabet:\n \"abcdefghi\\\n jklmnopqr\\\n stuvwxyz\"\n" gives abcdefghi jklmnopqr stuvwxyz; PyYAML gives abcdefghijklmnopqrstuvwxyz.
  • Cause yaml.tcl ::yaml::_parseDoubleQuoted folds line breaks before handling escapes, so an escaped line break (\ at end of line) becomes a space.
  • Prototype prototypes/800c0dc7c1.patch (44 lines, tests yaml-37.1, yaml-37.2). Removes \+newline+indentation before folding, keeping \\ pairs intact. Matches PyYAML on these cases.
  • Suggested action apply prototype
  • Suggested ticket comment Escaped line breaks in double-quoted scalars were folded into spaces. Proposed fix with tests yaml-37.*.
🔧af927e957f2022LDAP STARTTLS broken with Microsoft ADpatchREPRODUCED (9.1, 8.6). The 2022 fix 495ba2b8f5 made things worse: starttls and info whoami now always fail with `invalid command name "asnEnc…

Open ticket af927e957f · Bug · ldap · filed 2022-02-17

🔧 Since triage (updated 2026-11-16)

Status now: Open (Fixed)

  • 🔧 2026-11-15 check-in 1f10221741 on ldap-starttls-asncontext, not merged: Fix af927e957f: ldap::starttls and ldap::info whoami always failed with 'invalid command name' since tcllib 1.21. 495ba2b8f5 called asnEncodeString, which asn
  • 💬 2026-11-15 serhiy.storchaka: The fix in 495ba2b8f5 calls asnEncodeString, which asn does not export, so ldap::starttls and ldap::info whoami fail with "invalid command name" in tcllib 1.21, 2.0 and trunk, as noted above. Proposed fix in branch ldap-starttls-asncontext: use the exported asnContext 0 $oid, which encodes the same bytes. Test ldap-30.0 runs against a local mock LDAP server.
Full verdict

REPRODUCED (9.1, 8.6). The 2022 fix 495ba2b8f5 made things worse: starttls and info whoami now always fail with invalid command name "asnEncodeString". This affects tcllib 1.21, 2.0 and trunk.

  • Module ldap 1.10.2, asn 0.8.5
  • Tested in the ldap namespace, asnEncodeString is not found, because ::asn::asnEncodeString is not in asn's export list. Wrote a mock-server test with ldap::info whoami: it fails on trunk with that error on 9.1 and 8.6. With the fix the full ldap module passes 186/186 on 9.1 and 8.6.
  • Cause ldap.tcl ldap::Whoami and ldap::starttls call the unqualified, unexported asnEncodeString 80 .... The exported asnContext 0 $oid produces the same bytes (tag 0x80 = context-specific primitive [0]).
  • Prototype prototypes/af927e957f.patch (103 lines, test ldap-30.0. It checks the exact request bytes 7719 80 17 ... and the returned authzId)
  • Suggested action apply the prototype. Alternatively export asnEncodeString from asn and raise ldap's package require asn, as the reporter suggested. The prototype needs no asn change.
  • Suggested ticket comment 495ba2b8f5 calls asnEncodeString, which asn does not export, so starttls and whoami fail with "invalid command name" in 1.21/2.0. Proposed fix: use the exported asnContext 0 $oid (same encoding), with a mock-server test.
⬜2eb11228fb2021ftp::Put and ftp::Append don't handle -data {} properpatchREPRODUCED (9.1, 8.6)

Open ticket 2eb11228fb · Bug · ftp · filed 2021-07-10

Full verdict

REPRODUCED (9.1, 8.6)

  • Module ftp 2.4.14
  • Tested ftp::Put $h -data {} put.txt returns 1 but uploads the *string* "put.txt" to the remote file ftp.tmp. put.txt keeps its old contents. Append behaves the same. The attached ftp.patch is not in the local repo, so the fix was written from the code.
  • Cause ftp.tcl ::ftp::Put/::ftp::Append argument loops use $source == "" to decide whether a positional argument was already seen, so an empty -data value is overwritten by the remote name.
  • Prototype prototypes/2eb11228fb.patch (ftp.tcl 18 changed lines, counts the positional args; tests ftp-2.0/2.1/2.2 in both modes)
  • Suggested action compare with the attached patch on the web, apply, close as Fixed
  • Suggested ticket comment Confirmed: with -data {} the remote name was taken as the data and stored in ftp.tmp. Proposed fix with tests in the prototype.
🔧9833d8dca12021Can't change cron within a procpatchREPRODUCED (9.1, 8.6): cron::every test 10 mycron called from inside the task leaves the frequency at the old value. A one-shot cron::in task t…

Open ticket 9833d8dca1 · Bug · cron · filed 2021-06-11

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 28d29da86b on cron-task-reschedule, not merged: Fix 9833d8dca1: a cron task could not change its own schedule, because the command ran inside dict with and its changes were overwritten. Run the command outs
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch cron-task-reschedule: the task command ran inside dict with, which restored the old settings afterwards. It now runs outside, and a one-shot task that reschedules itself is kept. Tests cron-1.9 and cron-1.10.
Full verdict

REPRODUCED (9.1, 8.6): cron::every test 10 mycron called from inside the task leaves the frequency at the old value. A one-shot cron::in task that reschedules itself is deleted after it runs.

  • Module cron 2.2
  • Tested repro script plus sak test run modules/cron on 9.1 and 8.6
  • Cause cron.tcl runTasksCoro runs plain commands inside dict with processTable($task), which writes the old values back afterwards (as hypnotoad noted in 2024). One-shot tasks are also put on the cancel list before they run.
  • Prototype prototypes/9833d8dca1.patch (66 lines, tests cron-1.9, cron-1.10). Runs the command outside dict with and stores err/result afterwards. One-shot tasks are cancelled only if they were not rescheduled. running 0 is no longer written into a task that was deleted.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed. The task command runs inside dict with, which restores the old settings. Proposed patch runs it outside and keeps one-shot tasks that reschedule themselves.
🔧66b79317f32021SMTP - SASL::NTLM authentication failspatchREPRODUCED (9.1, 8.6)

Open ticket 66b79317f3 · Bug · sasl · filed 2021-04-29

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 554012e862 on sasl-ntlm-nonce-dash, not merged: Fix 66b79317f3: SASL NTLM authentication failed when the server nonce started with "-": DES::des took it for an option. Add -- to the DES::des calls.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch sasl-ntlm-nonce-dash: as the ticket suggests, -- is added to the DES::des calls, so a nonce starting with "-" isn't taken as an option. Test SASL-NTLM-1.2.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module sasl / SASL::NTLM 1.1.4
  • Tested SASL::NTLM::LMhash/NThash with a nonce starting with "-" (new test) on 9.1 and 8.6. Both fail with bad option "-...": must be one of -chunksize, ....
  • Cause sasl/ntlm.tcl: LMhash, NThash. DES::des ... -key $key $nonce has no --. A server nonce that starts with "-" is parsed as an option. DES::des already accepts --.
  • Prototype prototypes/66b79317f3.patch (62 lines, test SASL-NTLM-1.2). Adds -- to the three DES::des calls, as the ticket proposes. sasl tests: 21/21 pass on 9.1 and 8.6. The new test fails without the fix.
  • Suggested action apply prototype
  • Suggested ticket comment Still present on trunk. DES::des is called without --, so a server nonce that starts with "-" is taken as an option. Proposed fix with a test in prototypes/66b79317f3.patch.
⬜d5cf8839c32020[active mode] STOR data flushes before server connects to portpatchREPRODUCED (9.1, 8.6) with a strict mock server; pyftpdlib itself not available (no network)

Open ticket d5cf8839c3 · Bug · ftp · filed 2020-06-09

Full verdict

REPRODUCED (9.1, 8.6) with a strict mock server; pyftpdlib itself not available (no network)

  • Module ftp 2.4.14
  • Tested a server in a child process that, like pyftpdlib and ftpd, opens the active data connection right after PORT and delays the 200 reply. The client writes the whole upload and closes the data connection before it even sends STOR. A server that refuses data received before STOR (the mock answers 425) fails Put -data, Append -data and Put file. tcllib's own ftpd tolerates it, because it reads the already-buffered data later.
  • Cause ftp.tcl InitDataConn installs the writable handler (HandleOutput/HandleData) as soon as the data connection exists, before the server has accepted STOR/APPE. -buffering line is not the cause: full buffering only delays the same early write until the close.
  • Prototype prototypes/d5cf8839c3.patch (ftp.tcl ~48 changed lines: new StartPut, started on the 1xx reply to STOR/APPE or when the connection arrives later; error branches unset state.data in active mode too; CloseDataConn drops a pending writer. Tests ftp-4.0/4.1/4.2 with the strict server, plus the regular ftpd Get tests)
  • Suggested action apply prototype (it touches the same error branches as 1763707fff, so merge them together)
  • Suggested ticket comment The client starts sending as soon as the data connection is accepted, before the server has answered STOR; the buffering mode only changes when it is flushed. Proposed fix with tests in the attached prototype: data is sent only after the 1xx reply.
⬜7987932aaa2020Tcllib/httpd: Directory listing & error pages try to load /images/etoyoc-banner.jpg (non-configurable)patchREPRODUCED (9.1, 8.6)

Open ticket 7987932aaa · Bug · httpd · filed 2020-02-18

Full verdict

REPRODUCED (9.1, 8.6)

  • Module httpd 4.3.6
  • Tested GET on a directory without an index file through ::httpd::content.file. The listing contains <img src="/images/etoyoc-banner.jpg">. Tested with sak on 9.1 and 8.6.
  • Cause httpd.tcl / build/reply.tcl ::httpd::reply method html_header emits the author's site banner when no banner option is given. DirectoryListing and the default error templates call it without that option.
  • Prototype prototypes/7987932aaa.patch (65 lines, test httpd-client-0011). It emits <img> only when banner is given, and changes both the amalgamated httpd.tcl and build/reply.tcl. The test fails without the fix (1 1 instead of 1 0). Module tests pass on 9.1 and 8.6 (27/27).
  • Suggested action apply prototype
  • Suggested ticket comment Still present in 4.3.6: html_header falls back to a hard-coded /images/etoyoc-banner.jpg. Proposed fix: emit the banner only when the banner option is given (patch + test httpd-client-0011).
⬜8541c94be02019TeX/LaTeX syntax of diacritical marks in Bibtex fields is lost after parsingREPRODUCED (9.1, 8.6)

Open ticket 8541c94be0 · Bug · bibtex · filed 2019-09-12

Full verdict

REPRODUCED (9.1, 8.6)

  • Module bibtex 0.8
  • Tested Rapha{\"{e}}lle -> Rapha"elle, {{Oxytocin}} -> Oxytocin
  • Cause bibtex.tcl:ParseBibString drops every brace at any nesting depth and treats \ as an escape that is removed. Inner braces and TeX commands are lost.
  • Prototype none. Keeping inner braces/backslashes changes the values that existing users get, so it needs an opt-in option (e.g. -raw) and a decision.
  • Suggested action keep open; decide between an opt-in raw mode and a behaviour change
  • Suggested ticket comment Confirmed. ParseBibString strips all nested braces and backslashes. Preserving them would change results for existing users, so it would need a new option.
⬜e4d95d4ea02019TEPAM: using both -default and -multiple always includes defaultpatchREPRODUCED (9.1, 8.6)

Open ticket e4d95d4ea0 · Bug · tepam · filed 2019-07-31

Full verdict

REPRODUCED (9.1, 8.6)

  • Module tepam 0.5.4
  • Tested the ticket's example plus an unnamed -multiple variant (-named_arguments_first 0). mytest -myarg a gives mydefault a. The full tepam test suite passes with and without the patch on 9.1 (non-Tk files) and 8.6 (all files, Tk ones included).
  • Cause tepam.tcl, ProcedureArgumentEvaluation: the argument variable is first set to the default, then each provided value is lappended to it, for named arguments and for leading unnamed -multiple arguments. The named-arguments-first unnamed path already uses set ... [lrange ...] and is correct.
  • Prototype prototypes/e4d95d4ea0.patch (82 lines, tests tepam-bug_fixes.bug-e4d95d4ea0.1–5). The first provided value replaces the default (set), later ones are appended. Tests .2, .3 and .5 fail without the fix.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: provided values were appended to the default. A fix in which the first provided value replaces the default is prepared, with tests.
⬜7c8f294f7f2019clay: Duplicate tip479 tests ?patchREPRODUCED (9.1, 8.6): still 4x tip479-001..004 on trunk. Two sets are real duplicate names (the method variant and the Ensemble variant both u…

Open ticket 7c8f294f7f · Bug · A Category Is Missing · filed 2019-06-20

Full verdict

REPRODUCED (9.1, 8.6): still 4x tip479-001..004 on trunk. Two sets are real duplicate names (the method variant and the Ensemble variant both use tip479-001..004). The other two copies come from the intended "Repeating tests with 8.7 features" block that build.tcl appends. All 8 titles say "a later ensemble definition trumps a more primitive one", which does not match what the tests do (dictargs argument handling).

  • Module clay 0.8.8 (modules/clay/clay.test, generated from build/test.tcl)
  • Tested sak test run modules/clay on 9.1 and 8.6, before and after: 443/443 pass both times
  • Cause build/test.tcl:2504-2580 copies the tip479class block for tip479classE without renumbering
  • Prototype prototypes/7c8f294f7f.patch (24+/24- lines in build/test.tcl and clay.test). Renumbers the Ensemble variants to tip479-005..008 and gives all 8 tests titles that match what they test. No separate regression test, because the patch only changes tests.
  • Suggested action apply prototype, close as Fixed
  • Suggested ticket comment The method and Ensemble variants shared the names tip479-001..004 and had a copy-pasted title. Renumbered the Ensemble ones to 005..008 and gave them real titles. The 4x count comes from the 8.7 repeat block, which is intentional.
⬜b1c2d071482019ip::version and ip::mask accepts invalid netmask and other input after /patchREPRODUCED (9.1, 8.6)

Open ticket b1c2d07148 · Bug · ip · filed 2019-06-13

Full verdict

REPRODUCED (9.1, 8.6)

  • Module ip 1.5.1 (modules/dns/ip.tcl)
  • Tested ip::version 10.0.0.1/64 gives 4. ip::version 10.0.0.1/dsffds/44 gives 4. ip::mask 10.0.0.1/dsffds/44 gives dsffds. ip::prefix 10.0.0.1/64 errors with "negative shift argument". The attached diff is not retrievable (fossil artifact cannot resolve 927b8447…).
  • Cause ip.tcl:version splits on "/" and only checks the address part.
  • Prototype prototypes/b1c2d07148.patch (42 lines, tests ip-1.mask.1–16). version returns -1 for more than one "/", an empty mask, a bit count above 32/128, or a non-numeric mask that is not an address of the same family. ip::mask is unchanged (making it return -1 would change the API). dns module tests pass on 9.1 and 8.6.
  • Suggested action apply prototype. Decide separately whether ip::mask should validate.
  • Suggested ticket comment Confirmed on trunk. The attached diff is no longer retrievable from the repository. A prototype that makes ip::version validate the mask part is ready.
🔧4c3e5325002019tool::uuid_seed may raise errorpatchREPRODUCED (9.1, 8.6)

Open ticket 4c3e532500 · Bug · tool · filed 2019-02-20

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in ebee7660c7 on tool-uuid-seed-no-user, not merged: Fix 4c3e532500: package require tool failed when none of USERNAME, USER or user was set in the environment. tool::uuid_seed read env(user) unconditionally; dr
  • 💬 2026-11-15 serhiy.storchaka: Proposed fix in branch tool-uuid-seed-no-user: as pointed out in the ticket, the user name lookup is useless because its value is never used, so it is removed. Test tool-uuid-seed-noenv.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module tool 0.8 (tool.tcl, build/uuid.tcl)
  • Tested env -u USER -u USERNAME -u user tclsh with package require tool gives can't read "::env(user)": no such variable on 9.1 and 8.6. The package fails to load because uuid_seed runs at load time.
  • Cause tool.tcl and build/uuid.tcl, ::tool::uuid_seed: the code sets a local user from env(USERNAME), env(USER) or env(user) without checking the last one, and never uses it. The seed uses the literal string user@.
  • Prototype prototypes/4c3e532500.patch (81 lines, test tool-uuid-seed-noenv). It removes the dead lookup in both tool.tcl and build/uuid.tcl. The tool tests pass on 9.1 and 8.6 (187/187); without the fix the new test fails (1 of 187).
  • Suggested action apply prototype
  • Suggested ticket comment Still loads with an error when none of USERNAME/USER/user is set. The lookup is dead code (the result is never used), so it can just be removed; proposed patch with a test.
⬜333ccac7632018huddle wrap no longer accepts 2 argumentspatchREPRODUCED (9.1, 8.6). huddle wrap s abc gives "wrong # args: should be "huddle wrap node"", while huddle.man still documents only `huddle wrap t…

Open ticket 333ccac763 · Bug · huddle · filed 2018-12-06

Full verdict

REPRODUCED (9.1, 8.6). huddle wrap s abc gives "wrong # args: should be "huddle wrap node"", while huddle.man still documents only huddle wrap tag src. The man page's type-callback example (huddle to_node, huddle wrap D ..., setting) also describes the 0.1 API.

  • Module yaml/huddle 0.5
  • Tested 9.1 and 8.6 on trunk. 1.17 accepts the 2-argument form, 1.18 and later do not.
  • Cause huddle.tcl ::huddle::wrap {node} was changed to one argument in the 0.2 rewrite.
  • Prototype prototypes/333ccac763.patch (85 lines, test huddle-3.8). wrap accepts both node and the 0.1 form tag src (an empty tag gives HUDDLE src, as in 0.1), and huddle.man documents both forms. The test fails without the fix. Yaml module tests pass on 9.1 and 8.6.
  • Suggested action apply prototype, or fix only the docs if the 0.1 form should stay dropped. The outdated type-callback section of huddle.man needs a separate rewrite.
  • Suggested ticket comment Confirmed: the docs still show the 0.1 two-argument form. Proposed patch accepts both forms and documents them (test huddle-3.8).
⬜9a216372732018nettool cannot parse the ifconfig outputpatchREPRODUCED (9.1, 8.6)

Open ticket 9a21637273 · Bug · nettool · filed 2018-01-23

Full verdict

REPRODUCED (9.1, 8.6)

  • Module nettool 0.5.4
  • Tested nettool::dump on this Linux box (net-tools 2.x output format). It returns garbage like lo {ipv4 {{} 255.0.0.0} ...}, with no addresses and no MACs. broadcast_list gives only 127.0.0.1. hwid_list falls back to the fake 0x010203040506. ip_list fails with invalid command name "ldelete".
  • Cause the Linux ::nettool::dump only understands the old inet addr:X Bcast:Y Mask:Z / HWaddr format. The new format uses inet X netmask M broadcast B, ether MAC, and iface: with a trailing colon. Also ldelete, used by every ip_list, is never defined.
  • Prototype prototypes/9a21637273.patch (268 lines; about 50 lines of code, the same in nettool.tcl and build/; tests ifconfig-parse-old, ifconfig-parse-new, ip-list-ldelete). The parser is split out into ::nettool::ifconfig_parse and accepts both formats. It maps the new format to the old keys (addr:, Mask:, Bcast:, ether:), so callers are unchanged. The patch also defines ::ldelete next to ::ladd. With it, dump/ip_list/mac_list/hwid_list/broadcast_list return real data here. The new tests fail without the fix. 28/28 pass on 9.1 and 8.6. The attached nettool.diff (switch to ip address show and /proc/net/arp) is not in the local repo (blob missing), so I could not compare it.
  • Suggested action apply the prototype. Switching to ip/proc as the ticket suggests is a possible later step, because net-tools is deprecated.
  • Suggested ticket comment Confirmed on trunk: dump does not understand the net-tools 2.x ifconfig format, and ip_list also fails on the undefined ldelete. Proposed patch parses both formats and adds the missing helper, with tests.
⬜d74e6561a22017YAML->HUDDLE->YAML failspatchREPRODUCED (9.1, 8.6); the 2021 fix sits on unmerged branch tkt-d74e6561a2-yaml-huddle (f4046fd6a8)

Open ticket d74e6561a2 · Bug · yaml · filed 2017-09-28

Full verdict

REPRODUCED (9.1, 8.6); the 2021 fix sits on unmerged branch tkt-d74e6561a2-yaml-huddle (f4046fd6a8)

  • Module yaml 0.4.2
  • Tested yaml::huddle2yaml [yaml::yaml2huddle "a: 1\n..."] prints only --- and then the raw HUDDLE {!!map ...} text.
  • Cause yaml.tcl ::yaml::_imp_huddle2yaml knows only the huddle types string/list/dict, not yaml's mapping/sequence/str/int/…, so everything falls through to the default branch. Huddle number/boolean values are also emitted raw.
  • Prototype prototypes/d74e6561a2.patch (102 lines, tests yaml-32.1, yaml-32.2). Handles mapping/sequence like dict/list, dumps all scalar types (yaml and huddle number/boolean) as scalars. Same idea as the branch, but rewritten cleanly against trunk.
  • Suggested action apply prototype (or merge the branch after adding tests); then close the stale branch
  • Suggested ticket comment The fix on branch tkt-d74e6561a2-yaml-huddle was never merged. Proposed patch with tests yaml-32.* makes huddle2yaml handle yaml's own types.
⬜cd4fe2d0472017coroutine::auto causes auto_load to failpatchREPRODUCED (9.1, 8.6)

Open ticket cd4fe2d047 · Bug · (unused) · filed 2017-08-16

Full verdict

REPRODUCED (9.1, 8.6)

  • Module coroutine::auto 1.3
  • Tested reporter's script (clock format inside a coroutine) with coroutine::auto loaded. 9.1: invalid command name "::tcl::clock::SetupTimeZone" (auto_load); 8.6: can't upvar from variable to itself from global env in msgcat::Init
  • Cause coro_auto.tcl:wrap_global redirects every global inside a coroutine to coroutine::util::global (upvar #1). This includes Tcl's own globals (auto_index, auto_path, env, tcl_*), so auto_load, msgcat, etc. lose them. Qualified names (global ::foo) are mapped wrongly too.
  • Prototype prototypes/cd4fe2d047.patch (143-line patch: ~23 code lines plus a new coro_auto.test; tests coro-auto-global-1.0..1.3). Qualified names and Tcl-owned names (env, errorInfo, errorCode, argc, argv, argv0, auto_*, tcl_*, tk_*) go to the real global. Other names stay coroutine-local (1.2 checks this). 3 of the 4 tests fail without the fix. The patch adds the test file; the module had no tests.
  • Suggested action apply prototype, or decide on aku's 2017 idea to drop the global override entirely (a design decision)
  • Suggested ticket comment Still reproducible on 8.6 and 9.1. wrap_global also redirects Tcl's own globals (auto_index, auto_path, env) to the coroutine frame, which breaks auto_load and msgcat. A proposed fix routes qualified and Tcl-owned names to the real global, with a new test file.
⬜387599b3d82016csv::split -alternate doesn't work as expectedpatchREPRODUCED (9.1, 8.6): csv::split -alternate {a;b""b;c} \; gives a b {;c}, expected a b\"b c.

Open ticket 387599b3d8 · Bug · csv · filed 2016-07-27

Full verdict

REPRODUCED (9.1, 8.6): csv::split -alternate {a;b""b;c} \; gives a b {;c}, expected a b\"b c.

  • Module csv 0.10
  • Tested tclsh9.1, tclsh8.6; csv.test
  • Cause csv.tcl ::csv::Split, alternate DFA. A " in the middle of an unquoted value switches to the quoted-value state. The following " then closes the value, and the token after it is dropped.
  • Prototype prototypes/387599b3d8.patch (31 lines, tests csv-92.2.1–92.2.3). Adds a state for "" inside unquoted values. Lone quotes are handled as before.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed. A doubled quote inside an unquoted value confused the -alternate parser. Proposed fix treats "" there as a literal quote, as the regular parser and the documented format do.
🔧6bef0b5d0f2016bibtex package doesn't parse entries correctlypatchREPRODUCED (9.1, 8.6)

Open ticket 6bef0b5d0f · Bug · bibtex · filed 2016-05-15

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 68f4c9130b on bibtex-comment-anchor, not merged: Fix 6bef0b5d0f: bibtex silently dropped a record that contained the word "comment" or "preamble" in a field. Anchor the record-type regexps.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch bibtex-comment-anchor: the comment and preamble regexps are anchored like the string one, so a record containing those words is no longer dropped. Tests bibtex-7.0 to 7.2.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module bibtex 0.8
  • Tested a record with title = {No comment on this} is silently dropped (taken as @comment). note = {See preamble {x} here} is returned as a preamble.
  • Cause modules/bibtex/bibtex.tcl:ParseRecords. The comment and preamble regexps are not anchored with ^. The string one was fixed in 1.17. The attached patch is not in the repository (its artifact is missing).
  • Prototype prototypes/6bef0b5d0f.patch (69 lines, tests bibtex-7.0, 7.1, 7.2). It adds ^ to both regexps. 7.0 and 7.1 fail without the fix. All 24 tests pass on 9.1 and 8.6.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed. The comment/preamble regexps match anywhere in the entry. Fix in the prototype: anchor them like the string one.
  • Note a separate small bug: the comment regexp ([^\n])* captures only the last character, so -commentcommand gets e.g. } instead of the comment text.
⬜f990aa37bc2016oo::classmethod inherited only oncepatchREPRODUCED (9.1, 8.6)

Open ticket f990aa37bc · Bug · oo::util · filed 2016-02-22

Full verdict

REPRODUCED (9.1, 8.6)

  • Module oo::util 1.2.3
  • Tested the ticket script on 9.1 and 8.6: MySubclass clm works, MySubSubclass clm → unknown method. On 9.x the oo::util version replaces the native oo::define::classmethod (TIP 478), which does not have this bug.
  • Cause ooutil.tcl oo::class.Delegate create mixes in only the delegates of the direct superclasses. The delegates do not inherit from each other, so grandparent class methods are not seen.
  • Prototype prototypes/f990aa37bc.patch (25 changed lines, test ooutil-ticket-f990aa37bc). Each new delegate class now gets the superclasses' delegates as its superclasses, so the inheritance is transitive. Overriding and next also work (A/B/C chain checked). The test fails without the fix on 9.1 and 8.6. Conflicts textually with 3609183fff.patch, see there.
  • Suggested action apply prototype. Alternatively, on 8.7+/9, use the native classmethod instead of overriding it.
  • Suggested ticket comment Confirmed on 8.6 and 9.1: the class delegate only looked at direct superclasses. A proposed fix (delegates now inherit from the superclass delegates), with a test, is ready.
🔧dbde846f532015imap4 failspatchREPRODUCED (9.1, 8.6). Loading fails when argv0 does not exist (embedders like eggdrop) or when package require imap4 runs inside a proc.

Open ticket dbde846f53 · Bug · imap4 · filed 2015-12-21

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 799e94464d on imap4-load-without-argv0, not merged: Fix dbde846f53: package require imap4 failed when argv0 did not exist, e.g. in an embedded interpreter. Check for ::argv0 before comparing it with the script
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch imap4-load-without-argv0: the check is now [info exists ::argv0] && [info script] eq $::argv0. Adds imap4.test, which loads the package without argv0 and from inside a proc.
Full verdict

REPRODUCED (9.1, 8.6). Loading fails when argv0 does not exist (embedders like eggdrop) or when package require imap4 runs inside a proc.

  • Module imap4 0.5.5
  • Tested unset ::argv0; package require imap4 gives can't read "argv0": no such variable on 9.1 and 8.6.
  • Cause imap4.tcl, end of file: the demo client block if {[info script] eq $argv0} uses an unqualified argv0 (local scope inside a proc) and does not check that it exists. The reporter's info exists argv0 attempt failed because it was not namespace-qualified.
  • Prototype prototypes/dbde846f53.patch (71 lines, tests imap4-1.0 and imap4-1.1 in a new modules/imap4/imap4.test; the module had no tests). The condition becomes [info exists ::argv0] && [info script] eq $::argv0. Both tests fail without the fix and pass on 9.1 and 8.6. Moving the demo to examples/, as aku suggested, would be the larger alternative.
  • Suggested action apply prototype (the new test file needs fossil add)
  • Suggested ticket comment Still present in 0.5.5. Proposed fix uses ::argv0 guarded by info exists, with a new imap4.test covering loading without argv0 and from inside a proc.
⬜e048d944272015::term::ansi::ctrl::unix terminal dimensions functions work incorrectlypatchREPRODUCED (9.1, 8.6)

Open ticket e048d94427 · Bug · term · filed 2015-12-10

Full verdict

REPRODUCED (9.1, 8.6)

  • Module term::ansi::ctrl::unix 0.1.2
  • Tested ran columns/rows under util-linux script(1) with stty cols 123 rows 45 (ncurses 6.6 tput). With stdin on the tty: 123 45. With stdin a pipe or /dev/null (the reporter's echo ... | tclsh case): 80 24.
  • Cause ansi/ctrlunix.tcl columns/rows: exec tput cols <@stdin captures tput's stdout and stderr, so tput sees a terminal only through stdin. When stdin is not a tty it falls back to 80x24.
  • Prototype prototypes/e048d94427.patch (89 lines, new modules/term/ctrlunix.test with tests term-ctrl-unix-1.0 and 1.1, constraint: Linux with script/stty/tput). Adds 2>@stderr so tput can use the terminal on stderr. Test 1.1 fails without the fix. Side effect: a tput error message now goes to stderr instead of raising a Tcl error.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: when stdin is not a terminal, tput only sees pipes and reports 80x24. A fix that passes stderr through to tput is prepared, with a test.
🔧187a6fa7ae2015DNS client not handling partial TCP responses correctlypatchREPRODUCED (9.1, 8.6): if the first TCP read returns 1 byte, state(size) is never set and the query fails with an "Event error". The same happens…

Open ticket 187a6fa7ae · Patch · dns · filed 2015-11-16

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 79c8191d74 on dns-tcp-partial-reads, not merged: Fix 187a6fa7ae, 022bca0ff5: the dns client did not handle TCP replies that arrive in several reads. Buffer until the 2-byte length has arrived, read the len
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch dns-tcp-partial-reads, along the lines of the attached patch: buffer until the 2-byte length has arrived, read the length unsigned, and take the id from the reply. Test dns-4.1 uses a local server that splits the reply. 022bca0ff5 reports the same bug.
Full verdict

REPRODUCED (9.1, 8.6): if the first TCP read returns 1 byte, state(size) is never set and the query fails with an "Event error". The same happens after an empty read. The id check also scans the latest chunk instead of the reply, and a length above 32767 reads as negative.

  • Module dns 1.6.1
  • Tested local TCP server that sends the reply as 1 byte + 10 bytes + rest; dns tests on 9.1 and 8.6
  • Cause dns.tcl ::dns::TcpEvent
  • Prototype prototypes/187a6fa7ae.patch (64 lines, test dns-4.1). A small rewrite in the spirit of the attached patch: it buffers until 2 length bytes have arrived, takes the length unsigned and scans the id from the reply.
  • Suggested action apply prototype, then close 022bca0ff5 as a duplicate
  • Suggested ticket comment Confirmed with a local server that splits the reply. Proposed fix buffers until the 2-byte length has arrived, with a regression test. 022bca0ff5 reports the same bug.
🔧4811399edc2015comm disconnect after race conditionpatchREPRODUCED (9.1, 8.6)

Open ticket 4811399edc · Bug · comm · filed 2015-11-09

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 09fc83dd5b on comm-lost-conn-peer, not merged: Fix 4811399edc, 1491894fff: comm raised an error when a client disconnected after the "commIncoming race condition" warning. commLostConn now clears the pen
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch comm-lost-conn-peer: commLostConn assumed one connection per remote id. Now only the connection registered as the peer clears the pending sends and the peer entry, and the race warning is no longer printed for remote id 0, which all non-listening clients share. Test comm-4811399edc. Also fixes 1491894fff.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module comm 4.7.3
  • Tested the ticket's server/client scripts on 9.1 and 8.6 (server prints the race warning, then bgerror "can't read comm(::comm::comm,pending,0)"); new test fails unpatched with the same bgerror, passes patched; comm.test 23/24 pass (1 tls skip) on both, unchanged otherwise
  • Cause comm.tcl commLostConn: when two connections share a remote id, only the first is registered in peers/pending (commNewConn), but losing either one unsets pending,$id, so the second loss fails. Also commIncomingOffered warns about a "race" for remote id 0, which non-listening peers always share
  • Prototype prototypes/4811399edc.patch (138 lines incl. reindent of a 25-line block, test comm-4811399edc): commLostConn processes pending/peers only if the lost fid is the registered peer; no race warning for remote id 0
  • Suggested action apply prototype
  • Suggested ticket comment commLostConn assumed one connection per remote id. Proposed fix in a branch: only the connection registered as the peer cleans up pending sends, and the warning is dropped for non-listening peers (id 0); test comm-4811399edc.
⬜13d8e3ca062015websocket serve: clean close handshakepatchREPRODUCED (9.1, 8.6). websocket::close never sends the close frame: the peer receives nothing and the socket is just closed.

Open ticket 13d8e3ca06 · Patch · websocket · filed 2015-10-22

Full verdict

REPRODUCED (9.1, 8.6). websocket::close never sends the close frame: the peer receives nothing and the socket is just closed.

  • Module websocket 1.6
  • Tested 127.0.0.1 socket pair, websocket::takeover on the server side, websocket::close $srv 1000. The client read 0 bytes.
  • Cause websocket.tcl ::websocket::close sets Connection(state) CLOSED before calling send $sock 8, and send refuses to write unless the state is CONNECTED. The second half of the ticket (waiting for the peer's close reply instead of disconnecting at once) is a bigger change and is not in the prototype.
  • Prototype prototypes/13d8e3ca06.patch (106 lines, test websocket-close-1.0, helper wsLocalPair). close now sets a transient state CLOSING, and send allows only opcode 8 in that state. The early return now also covers CLOSING. That prevents recursion when the close frame cannot be written: send fails, calls close 1001, and that nested call returns immediately; I checked this with a peer that had already gone. The new test passes on 9.1 and 8.6 and fails without the fix. I ran it locally, without the network tests.
  • Suggested action apply prototype. Keep open, or file separately, for waiting on the close handshake.
  • Suggested ticket comment Confirmed in 1.6: the state is set to CLOSED before the close frame is sent, so the frame is never sent. Proposed patch sends it (test included); waiting for the peer's close reply is still open.
⬜d8795764382015nettool ifconfig 'couldn't execute "ifconfig": no such file or directory'patchREPRODUCED (9.1, 8.6)

Open ticket d879576438 · Bug · nettool · filed 2015-06-22

Full verdict

REPRODUCED (9.1, 8.6)

  • Module nettool 0.5.4 (modules/nettool/nettool.tcl, amalgamated from build/*.tcl)
  • Tested PATH=/usr/bin:/bin, then nettool::dump/ip_list/hwid_list/mac_list/broadcast_list/arp_table on 9.1 and 8.6. All fail with couldn't execute "ifconfig" (or "arp"). ifconfig/arp are in /usr/sbin here. Package loaded with the pkgIndex typo from 61e49798ee fixed locally.
  • Cause nettool.tcl ::nettool::dump and ::nettool::arp_table (Linux and macOS variants) run exec ifconfig and exec arp -a without a path. /sbin and /usr/sbin are often not in PATH for ordinary users.
  • Prototype prototypes/d879576438.patch (228 lines, about 60 of them code: same change in nettool.tcl and in build/core.tcl + build/platform_unix*.tcl; test dump-sbin-path). Adds ::nettool::sbin_exec: it uses auto_execok and falls back to /sbin, /usr/sbin, /usr/local/sbin. The 4 exec calls now go through it. The test drops sbin from PATH and calls dump. It fails without the fix. The module's tests pass 26/26 on 9.1 and 8.6.
  • Suggested action apply the prototype (with or after 9a21637273, which touches the same dump proc).
  • Suggested ticket comment Still fails on trunk when /sbin and /usr/sbin are not in PATH. Proposed fix: look up ifconfig/arp with auto_execok and fall back to the sbin directories. Patch with test available.
⬜631a44ed662015WebSocket clients can block serverREPRODUCED (9.1, 8.6). A peer that sends 1 byte and stalls freezes the server's event loop until it closes. In my test, a 100 ms after canary tic…

Open ticket 631a44ed66 · Bug · websocket · filed 2015-04-16

Full verdict

REPRODUCED (9.1, 8.6). A peer that sends 1 byte and stalls freezes the server's event loop until it closes. In my test, a 100 ms after canary ticked twice in about 2 s instead of about 20 times.

  • Module websocket 1.6
  • Tested local 127.0.0.1 server with websocket::takeover; a second tclsh sends 1 byte, sleeps 2 s, closes.
  • Cause websocket.tcl ::websocket::takeover sets -blocking on, and ::websocket::Receiver does read $sock N for the header, length, mask and payload. Writes in send are blocking too.
  • Prototype none (needs a non-blocking, buffered rewrite of Receiver, or a coroutine; this is more than 60 lines)
  • Suggested action keep open
  • Suggested ticket comment Still reproducible in 1.6: the socket is blocking and Receiver reads whole frame parts at once. A fix needs a non-blocking Receiver (e.g. coroutine based).
🔧7446477f892015tar package: issues with long file namepatchREPRODUCED (9.1, 8.6)

Open ticket 7446477f89 · Bug · tar · filed 2015-01-28

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in a773e07f3c on tar-gnu-longlink, not merged: Fix 7446477f89: tar::create rejected names that do not fit into the prefix/name fields or are longer than 255 characters. Write a GNU ././@LongLink entry for
  • 🔧 2026-11-16 check-in 4b96e54440 on tar-gnu-longlink, not merged: Run tar-tkt-7446477f89 only on Unix: the names are longer than MAX_PATH on Windows, and the result is compared with the output of the system tar.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch tar-gnu-longlink: names that can't be split into the ustar prefix and name fields get a GNU ././@LongLink entry, which tar already reads. Test tar-tkt-7446477f89 checks the result with tar::contents and with the system tar tf. It runs only on Unix, because such names exceed MAX_PATH on Windows.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module tar 0.15
  • Tested tar::create with a 154-char name without /, and with a 271-char path. It fails with "path name cannot be split into prefix and name" or "path name over 255 chars". GNU tar handles both with a ././@LongLink (type L) entry, which tar.tcl already reads (HandleLongLink).
  • Cause tar.tcl formatHeader supports only the ustar prefix/name split. Fix: when the name cannot be split or exceeds 255 chars, emit a GNU @LongLink entry with the full name, followed by the normal header with a truncated name.
  • Prototype prototypes/7446477f89.patch (98 lines, test tar-tkt-7446477f89; checks both tar::contents and GNU tar tf). The new test fails without the fix. The tar module passes on 9.1 and 8.6 (13/13). Pre-existing limitation, not addressed: tar::remove does not understand LongLink entries.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: names that cannot be split into ustar prefix/name are rejected, while GNU tar writes a ././@LongLink entry for them. Since tar already reads that format, proposed fix: write it too (patch with test).
⬜665fbc24eb2014error in destructor hides error in constructorpatchREPRODUCED (9.1, 8.6; snit 1 and 2)

Open ticket 665fbc24eb · Bug · snit · filed 2014-12-19

Full verdict

REPRODUCED (9.1, 8.6; snit 1 and 2)

  • Module snit 2.3.4 / 1.4.3 (RT.type.typemethod.create, RT.widget.typemethod.create, RT.DestroyObject)
  • Tested the ticket script gives err=dtor error, and the constructor error is lost. Also, the half-destroyed object is left behind: the instance command still exists and dog destroy then fails with "Called 'destroy' method in constructor".
  • Cause the create typemethods call RT.DestroyObject uncaught after a constructor error. RT.DestroyObject aborts in the middle of its cleanup if the user destructor throws.
  • Prototype prototypes/665fbc24eb.patch (204 lines, tests error-1.3a (type) and error-1.3b (widget, Tk)). The create paths keep the constructor error and append the destructor's errorInfo. RT.DestroyObject finishes the cleanup (instance command, hull, namespace) even if the destructor fails, then rethrows. snit.test passes on tclsh9.1/8.6 and wish9.1/8.6 (Xvfb). Without the fix, error-1.3a/b fail, and on tclsh 82 later tests cascade because of the leftover object.
  • Suggested action apply prototype (review: $obj destroy with a failing destructor now really destroys the object before rethrowing)
  • Suggested ticket comment Confirmed. Proposed fix: the constructor error is reported, with the destructor error appended to errorInfo, and the object is fully cleaned up even when the destructor fails. Tests error-1.3a/b.
🔧e3a6e3e02a2014Huddle: creating a list with quote as element failspatchREPRODUCED (9.1, 8.6)

Open ticket e3a6e3e02a · Bug · huddle · filed 2014-08-25

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 4915d98a3c on huddle-isHuddle-nonlist, not merged: Fix e3a6e3e02a: huddle list with a quote as element failed, because isHuddle applied list commands to strings that are not lists. Check string is list first.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch huddle-isHuddle-nonlist: isHuddle checks string is list before applying list commands. Test huddle-3.7.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml/huddle 0.5
  • Tested huddle list \" raises "unmatched open quote in list" on trunk (9.1, 8.6) and in 1.17 and 1.18. huddle list \{ fails the same way.
  • Cause huddle.tcl ::huddle::isHuddle runs lindex $obj 0 on every argument before checking that it is a list.
  • Prototype prototypes/e3a6e3e02a.patch (41 lines, test huddle-3.7). The check is now ![string is list $obj] || [llength $obj] != 2 || [lindex $obj 0] ne "HUDDLE". The test fails without the fix. The yaml module tests pass on 9.1 and 8.6 (157/157).
  • Suggested action apply prototype
  • Suggested ticket comment Still fails: isHuddle applies list operations to arbitrary strings. Proposed fix checks string is list first (patch + test huddle-3.7).
⬜577ed0d10b2014Variables not unset unless ::websocket::upgrade not invokedREPRODUCED (9.1, 8.6). After a successful websocket::test without upgrade, ::websocket::Client_<srv>_<cli> stays set.

Open ticket 577ed0d10b · Bug · websocket · filed 2014-08-11

Full verdict

REPRODUCED (9.1, 8.6). After a successful websocket::test without upgrade, ::websocket::Client_<srv>_<cli> stays set.

  • Module websocket 1.6
  • Tested websocket::test s0 c0 /x {...} followed by info vars ::websocket::Client_*.
  • Cause websocket.tcl ::websocket::test creates the context; only ::websocket::upgrade unsets it.
  • Prototype none. This needs an API decision: either document the context and make the caller unset it (as the maintainer suggested in 2014), or add a websocket::discard cliSock.
  • Suggested action keep open (human decision on the API)
  • Suggested ticket comment Still the case in 1.6. The simplest option is probably to document that a context is kept until upgrade, and to add a small command to drop it when the server decides not to upgrade.
⬜bb5f70d7dd2014Client's sub-protocol preference not respectedREPRODUCED (9.1). Behaves as described; it is really an RFE. With handlers live /x h1 a and live /x h2 b, a client offering b, a gets a fro…

Open ticket bb5f70d7dd · Bug · websocket · filed 2014-08-11

Full verdict

REPRODUCED (9.1). Behaves as described; it is really an RFE. With handlers live /x h1 a and live /x h2 b, a client offering b, a gets a from h1: the order in which handlers were registered wins.

  • Module websocket 1.6
  • Tested websocket::test with two handlers, inspecting Client(protocol).
  • Cause websocket.tcl ::websocket::test loops over the server's handlers and takes the first one that matches any client protocol.
  • Prototype none. RFC 6455 lets the server choose, so changing the default order is an API/behaviour decision (iterate client protocols first, or add priorities).
  • Suggested action keep open as RFE
  • Suggested ticket comment Still the case: the order of handler registration decides. Iterating over the client's list first would honour its preference; that is a behaviour change that needs a decision.
⬜fefcf367882014remoteid-to-socket ambiguity in commREPRODUCED (9.1, 8.6; design limitation)

Open ticket fefcf36788 · Bug · comm · filed 2014-02-06

Full verdict

REPRODUCED (9.1, 8.6; design limitation)

  • Module comm 4.7.3
  • Tested same mechanism as 4811399edc (two connections with the same remote id): the second connection gets only a fids entry, peers,$id keeps the first socket; on disconnect commLostConn raised a bgerror
  • Cause comm.tcl commNewConn keeps only one socket per id; ids are {port host} as seen by the receiver, so tunnels/NAT make different peers collide
  • Prototype prototypes/4811399edc.patch fixes the error when one of the colliding connections is lost; the ambiguity itself (sending to the second peer) needs an id redesign
  • Suggested action keep open as RFE after applying the 4811399edc fix
  • Suggested ticket comment Ids are what the receiver sees ({port host}), so peers behind the same tunnel collide. The error on disconnect is fixed with 4811399edc; addressing colliding peers separately needs an id redesign.
⬜3613725fff2013::rest::get & "format" config option...patchREPRODUCED (9.1, 8.6): rest::get $url {} {format json} returns the raw string. Only result is honoured in rest::simple, although rest.man lis…

Open ticket 3613725fff · Bug · rest · filed 2013-05-22

Full verdict

REPRODUCED (9.1, 8.6): rest::get $url {} {format json} returns the raw string. Only result is honoured in rest::simple, although rest.man lists format for simple (and uses it in the Twitter example). create_interface accepts both.

  • Module rest 1.8
  • Tested local HTTP server on 127.0.0.1 returning JSON. format json returned raw text, result json returned a dict.
  • Cause rest.tcl:simple checks only dict exists $config result.
  • Prototype prototypes/3613725fff.patch (115 lines; adds a new modules/rest/rest.test with a local HTTP server; tests rest-1.0/1.1). It falls back to format like create_interface does. rest-1.0 fails without the fix and both pass on 9.1 and 8.6. The rest module had no test file before. The test stubs tdom if it is missing, since only JSON is used.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: rest::simple ignores format although the docs and create_interface treat it as a synonym of result. Proposed fix accepts both, with a test.
⬜3612359fff2013lexnormalize ignore leading ../patchREPRODUCED (9.1, 8.6): fileutil::lexnormalize ../toto gives toto; a/../../b gives b

Open ticket 3612359fff · Bug · fileutil · filed 2013-04-30

Full verdict

REPRODUCED (9.1, 8.6): fileutil::lexnormalize ../toto gives toto; a/../../b gives b

  • Module fileutil 1.16.4
  • Tested 9.1 and 8.6. The existing fu-lexnormalize-1.* tests (which even expect ../a -> a) never run, because they are guarded by the non-existent internal LexNormalize.
  • Cause fileutil.tcl lexnormalize drops .. for relative paths when nothing is left to remove. Note: fileutil::jail relies on the current behaviour, so the attached fileutil.tcl must not be applied as is; it is also wrong for absolute /...
  • Prototype prototypes/3612359fff.patch (+31/-3, tests fu-lexnormalize-2.0..9, jail-2.3.1). Leading .. are kept for relative paths (absolute paths unchanged). jail now strips leading .. itself, so its behaviour is unchanged (jail-2.3 and the new jail-2.3.1 pass). Without the fix, fu-lexnormalize-2.3..2.7 fail on 9.1 and 8.6. With it, fileutil passes 463/463 on both.
  • Suggested action apply the prototype (it changes documented-by-example behaviour for relative paths; mention it in changes)
  • Suggested ticket comment Confirmed. Proposed fix keeps leading ../ for relative paths. fileutil::jail depended on the old behaviour, so it now strips leading .. itself. Tests added.
⬜3609183fff2013ooutil::singleton not working when it is used as mixin classpatchREPRODUCED (9.1, 8.6)

Open ticket 3609183fff · Bug · oo::util · filed 2013-03-27

Full verdict

REPRODUCED (9.1, 8.6)

  • Module oo::util 1.2.3
  • Tested the doc example (oo::class create example {self mixin ooutil::singleton; ...}) returns two different objects on 9.1 and 8.6; ooutil::singleton create ... works. Applying the mixin after creation (oo::objdefine example mixin ooutil::singleton) also works.
  • Cause ooutil.tcl oo::class.Delegate create (classmethod support, added in the same release) runs the definition script, then calls oo::objdefine $cls mixin {*}$superdelegates. This replaces all object mixins, so the self mixin singleton set by the script is lost. The doc example also uses the unqualified name singleton, which does not resolve.
  • Prototype prototypes/3609183fff.patch (18 changed lines, test ooutil-ticket-3609183fff). It keeps the existing mixins (mixin {*}$superdelegates {*}[info object mixins $cls]) and changes the example in ooutil.man and the code comment to ooutil::singleton. The test fails without the fix on 9.1 and 8.6. Conflicts textually with f990aa37bc.patch (same line): if both are applied, the line becomes oo::objdefine $cls mixin $delegate {*}[info object mixins $cls].
  • Suggested action apply prototype
  • Suggested ticket comment Cause: the classmethod support in oo::util (oo::class.Delegate) overwrote the object mixins set by the class definition script. A proposed fix that keeps them, with a test, is ready; the doc example now uses ooutil::singleton.
⬜3584575fff2012Improved error messagesREPRODUCED (9.1, 8.6). Partly addressed by the 472259ffff prototype.

Open ticket 3584575fff · RFE · cmdline · filed 2012-11-06

Full verdict

REPRODUCED (9.1, 8.6). Partly addressed by the 472259ffff prototype.

  • Module cmdline 1.5.3 (typedGetoptions)
  • Tested typedGetoptions with -n abc for n.integer gives only the usage text. The specific "Option requires integer argument -- n" is dropped.
  • Cause typedGetoptions throws [typedUsage] and discards $arg. The attached patch is reversed (p4 "#7 - #1"), against 1.3.3, and bundles the file/dir and .must features plus a fileutil dependency.
  • Prototype none separate. prototypes/472259ffff.patch exposes the failed condition in the errorcode (CMDLINE USAGE INVALID {Option requires integer argument -- n}, test typed-cmdline-8.1).
  • Suggested action keep open; close together with 472259ffff if that prototype is applied
  • Suggested ticket comment The specific reason is still discarded. A proposed change for 472259ffff puts it into the errorcode. The attached patch is reversed and mixes in other features.
🔧3578420fff2012[exists] and colorful generator namespatchREPRODUCED (9.1, 8.6)

Open ticket 3578420fff · Bug · generator · filed 2012-10-19

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in e6d2fc9a61 on generator-exists-colorful-names, not merged: Fix 3578420fff: generator exists failed for generator names with glob or other special characters. Use namespace which -command. Add generator.test.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch generator-exists-colorful-names: exists uses namespace which -command instead of passing the name to info commands as a pattern. Adds generator.test.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module generator 0.3
  • Tested a coroutine named ::g[1] used with generator foreach gives no items on 9.1 and 8.6. generator exists ::g? returns 1 if a command ::gx exists.
  • Cause generator.tcl exists uses info commands $generator, which treats the name as a glob pattern.
  • Prototype prototypes/3578420fff.patch (72 lines; exists now uses namespace which -command; adds the module's first test file, generator.test, with tests generator-exists-1.0/1.1). Both tests fail without the fix and pass with it on 9.1 and 8.6.
  • Suggested action apply prototype
  • Suggested ticket comment Still present: exists passes the name to info commands as a pattern. Proposed fix using namespace which -command, with a new generator.test.
⬜3506188fff2012level none cannot be inheritedpatchREPRODUCED (9.1, 8.6)

Open ticket 3506188fff · Bug · logger · filed 2012-03-16

Full verdict

REPRODUCED (9.1, 8.6)

  • Module logger 0.9.5
  • Tested the ticket's script fails with Invalid level 'none' - levels are ... on trunk with 9.1 and 8.6. With the fix the child is created at level none and all its level commands are no-ops. The log module passes 372/372 on 9.1 and 8.6.
  • Cause in logger.tcl ::logger::init, the parent walk calls enable $enabled, and the tree setlevel calls disable/enable with the parent's currentloglevel. Both reject the pseudo-level none.
  • Prototype prototypes/3506188fff.patch (69 lines, test logger-18.0). The tree setlevel accepts none and disables every level. The parent walk skips enable for none. Side effect: ${log}::setlevel none now works too.
  • Suggested action apply the prototype.
  • Suggested ticket comment Reproduced on trunk: init of a child fails because none is passed to enable/setlevel. Proposed fix with test logger-18.0: setlevel treats none as "disable all", and the init walk skips it.
⬜3475081fff2012Parser accepts invalid YAML input, mistranslates otherspatchREPRODUCED (9.1, 8.6)

Open ticket 3475081fff · Bug · yaml · filed 2012-01-17

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested {list-no-comma: [ "a" "b" "c" ]} gives {list-no-comma {a b c}}; {x: [ "a": "b", "c": "d" ]} gives {x {a b c d}}. Compared with PyYAML: the first is an error, the second is a list of single-pair mappings.
  • Cause yaml.tcl ::yaml::_parseFlowNode ignores commas between flow entries. A : inside a flow sequence just adds another sequence element.
  • Prototype prototypes/3475081fff.patch (103 lines incl. 4 tests yaml-41.1–41.4, ~25 code lines). It adds a MISSING_COMMA error for flow sequences and mappings, and builds a single-pair mapping for k: v inside [...].
  • Suggested action apply prototype (needs review: rejecting comma-less input is stricter than before)
  • Suggested ticket comment Proposed fix in the flow parser: missing commas now raise an error, and [a: b, c: d] yields single-pair mappings as the spec requires; tests yaml-41.*.
⬜3434579fff2011Infinite looppatchREPRODUCED (9.1, 8.6)

Open ticket 3434579fff · Bug · yaml · filed 2011-11-07

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested yaml2dict "a: |\n x", "a: >\n x" and "a: x # c" (no final newline) hang on both versions. With a final newline they work.
  • Cause yaml.tcl ::yaml::_getLine sets data(start) to length+1 when the last line has no newline. _eof tests start == length and never becomes true, so _parseBlockScalar and the block parser loop forever.
  • Prototype prototypes/3434579fff.patch (61 lines, tests yaml-40.1, yaml-40.2). The tests run in a child interp with a command limit, so the unfixed code fails instead of hanging.
  • Suggested action apply prototype
  • Suggested ticket comment _getLine moved past the end of a buffer without a final newline, so the EOF check never fired. Proposed fix clamps the position; tests yaml-40.*.
⬜3312860fff2011Many FTP commands do not handle commands with absolute pathsREPRODUCED (9.1, 8.6): after CWD sub, absolute names such as SIZE /sub/f.txt are resolved against the current directory instead of the root.

Open ticket 3312860fff · Bug · ftpd · filed 2011-06-07

Full verdict

REPRODUCED (9.1, 8.6): after CWD sub, absolute names such as SIZE /sub/f.txt are resolved against the current directory instead of the root.

  • Module ftpd 1.4.1
  • Tested in-process ftpd with a raw client on 127.0.0.1.
  • Cause the commands strip the leading "/" before joining the name to the current directory.
  • Prototype a fix with tests exists and will be reviewed privately, because the path handling has security implications; the patch attached to the ticket should not be applied as is.
  • Suggested action review the prototype privately, then release.
  • Suggested ticket comment Confirmed; absolute names are resolved against the current directory instead of the root. A fix is being prepared.
🔧3308105fff2011Traceback fingers unrelated catch outputpatchREPRODUCED (9.1, 8.6; changed form)

Open ticket 3308105fff · Bug · smtp · filed 2011-05-26

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 025ab5c0dc on smtp-rejected-message-error, not merged: Fix 3308105fff: since tcllib 1.21 (1df85e69fb) a message rejected by the server made smtp::sendmessage return code 7 with a raw dict instead of raising an err
  • 💬 2026-11-15 serhiy.storchaka: Since tcllib 1.21 (1df85e69fb), when the server rejects a message, smtp::sendmessage returns code 7 with a raw dict instead of raising an error, and a rejected Bcc copy is not reported at all. Proposed fix in branch smtp-rejected-message-error, with a new smtp.test that uses a fake SMTP server.
Full verdict

REPRODUCED (9.1, 8.6; changed form)

  • Module mime / smtp 1.5.2
  • Tested local fake server that answers 451 after DATA. sendmessage now returns code 7 with the raw response dict (diagnostic {...} args {} code 451), not an error. With -Bcc, a rejected second (Bcc) copy is not reported at all.
  • Cause smtp.tcl: sendmessage. 1df85e69fb (2021, "Fix goofy error handling", in 1.21) replaced the old code-7-to-error conversion, which used a stale errorInfo (the original report), with return -options $copts $cres. That leaks the internal code 7, and copts2/cres2 of the Bcc send are dropped. Side note, not fixed: the header-restore code from 8340b8ed2f is dead because of typos (dict set origheaderx, dict get orignames without $).
  • Prototype prototypes/3308105fff.patch (174 lines, mostly new smtp.test; tests smtp-3.1, smtp-3.2). Turns code 7 back into a plain error "451: diagnostic" with fresh errorInfo, and propagates the Bcc result. mime tests: 104/104 on 9.1 and 8.6. Both new tests fail without the fix.
  • Suggested action apply prototype. Also look at the header-restore typos.
  • Suggested ticket comment The stale errorInfo is gone, but since 1.21 a rejected message makes sendmessage return code 7 with a raw dict instead of an error, and Bcc failures are lost. Proposed fix with tests in prototypes/3308105fff.patch.
⬜3164978fff2011missing procs or wrong manpatchREPRODUCED (9.1, 8.6)

Open ticket 3164978fff · Bug · nmea · filed 2011-01-24

Full verdict

REPRODUCED (9.1, 8.6)

  • Module nmea 1.1.0
  • Tested nmea.man documents ::nmea::close_port, ::nmea::close_file and ::nmea::rate. None of them exist (invalid command name). ::nmea::close and ::nmea::configure exist but are not documented. Also nmea::configure rate N with no open file fails (can't read nmea(fh)), although the man page says no file needs to be open.
  • Cause nmea.tcl: the API was renamed (close / configure) without updating the man page. In ::nmea::configure the rate branch schedules read_file on $nmea(fh) unconditionally.
  • Prototype prototypes/3164978fff.patch (139 lines, incl. new nmea.test with tests nmea-1.0, 2.0, 2.1, 3.0). Adds close_port, close_file and rate as thin wrappers, as the man page asks. Guards the configure rate branch with [info exists nmea(fh)]. Documents close and configure. nmea had no test file. 3 of the 4 new tests fail without the fix, and all pass on 9.1 and 8.6. Use fossil diff --new-file / fossil add for nmea.test.
  • Suggested action apply the prototype.
  • Suggested ticket comment Still true on trunk. Proposed patch adds the documented close_port/close_file/rate commands, documents close/configure, and fixes configure rate without an open file. It adds the first nmea tests.
⬜3127781fff2010term::interact::{menu,pager} consume 100% CPUpatchREPRODUCED (9.1, 8.6): the probable cause is reproduced, not the CPU load itself, because the reporter gave no script

Open ticket 3127781fff · Bug · term · filed 2010-12-05

Full verdict

REPRODUCED (9.1, 8.6): the probable cause is reproduced, not the CPU load itself, because the reporter gave no script

  • Module term::receive 0.2, term::interact::menu 0.2, term::interact::pager 0.2
  • Tested ran a menu interact on a chan pipe input (fed "\n"). Afterwards the input channel stays non-blocking (fconfigure -blocking = 0). After interact on stdin, any later gets stdin/read stdin loop in the application returns at once instead of waiting, which spins at 100% CPU.
  • Cause receive.tcl: term::receive::listen sets -blocking 0 and unlisten never restores it. menu/pager interact call listen/unlisten on -in (stdin by default).
  • Prototype prototypes/3127781fff.patch (148 lines, of which ~120 are a new test file modules/term/receive.test with tests term-receive-1.0, 1.1 and term-interact-menu-1.0). listen remembers the previous blocking mode per channel and unlisten restores it. The term module had no tests before. 1.0 and the menu test fail without the fix.
  • Suggested action apply prototype; keep the ticket open for confirmation if desired
  • Suggested ticket comment After interact, the input channel (stdin) was left non-blocking, so later reads of stdin spin instead of waiting. A fix that restores the blocking mode in unlisten is prepared, with tests.
⬜3107937fff2010Manual/Modification: when is autoproxy hooked to httppatchREPRODUCED (9.1, 8.6): configure -proxy_host alone still does not install http::config -proxyfilter, and the docs do not say so (they do not do…

Open ticket 3107937fff · Bug · http :: autoproxy · filed 2010-11-12

Full verdict

REPRODUCED (9.1, 8.6): configure -proxy_host alone still does not install http::config -proxyfilter, and the docs do not say so (they do not document init's arguments either)

  • Module autoproxy 1.8.1
  • Tested code reading. I also tried the attached patch idea (install the filter in configure) on top of the 29ff5c3abd test setup. It works, but then https through the tunnel sends GET https://host/... (absolute URL). That is an http package issue (Tcl 2d9b31130b; on Tcl 9 you need http::register https 443 ... {} 0 1). The tls tests also need a reset of -proxyfilter.
  • Cause autoproxy.tcl: only init calls http::config -proxyfilter. (The attached patch also has a bug: string length options(proxy_host) is missing the $.)
  • Prototype prototypes/3107937fff.patch (35 lines, docs only: documents init's httpproxy/no_proxy arguments, the return value, and that only init installs the filter). Docs only, so no test.
  • Suggested action apply the docs prototype; keep open or close depending on whether configure should install the filter (human decision, see the https interaction above)
  • Suggested ticket comment The docs now explain init's arguments and that only init installs the proxy filter. Installing it from configure would also affect https requests (Tcl http 2d9b31130b), so it needs a separate decision.
⬜3052576fff2010Lines starting with # in block scalars are ignoredpatchREPRODUCED (9.1, 8.6)

Open ticket 3052576fff · Bug · yaml · filed 2010-08-24

Full verdict

REPRODUCED (9.1, 8.6)

  • Module yaml 0.4.2
  • Tested "a: |\n line\n # mid\n end\n" gives a {line\nend\n}; the # line is dropped (only the first content line survives). Compared with PyYAML.
  • Cause yaml.tcl ::yaml::_parseBlockScalar skips every line starting with #, whatever its indentation.
  • Prototype prototypes/3052576fff.patch (49 lines, tests yaml-39.1, yaml-39.2). Only lines indented less than the content are treated as comments.
  • Suggested action apply prototype
  • Suggested ticket comment Lines at content indentation in a literal/folded block scalar were dropped as comments. Proposed fix: only less-indented # lines are treated as comments; test yaml-39.*.
⬜3025582fff2010comments lost, positions changingpatchREPRODUCED (9.1, 8.6) / PARTLY-FIXED (indented comments are kept since c443dd347c, 2014, ticket 3612465)

Open ticket 3025582fff · Bug · inifile · filed 2010-07-05

Full verdict

REPRODUCED (9.1, 8.6) / PARTLY-FIXED (indented comments are kept since c443dd347c, 2014, ticket 3612465)

  • Module inifile 0.3.3
  • Tested 9.1, 8.6. Opened and committed an ini file without changes. Section order comes out in hash order and keys in lsort -dictionary order. A comment after the last key of a section ends up above whatever section is written next. A trailing comment at end of file is dropped.
  • Cause ini.tcl:commit iterates array names sections and sorts keys. _loadfile drops pending comments at EOF. The comment for "after last key" is attached to the following section header, so it moves when sections are reordered.
  • Prototype prototypes/3025582fff.patch (97 lines, test inifile-6.0). Records a creation sequence number for sections (value of sections()) and for keys (new order array). commit writes in file/creation order. Trailing comments are kept in a trailer variable and written at the end. The return order of ini::sections/keys is unchanged (test inifile-1.2 depends on hash order).
  • Suggested action apply prototype. Optionally also return ini::sections/ini::keys in file order.
  • Suggested ticket comment Still reproducible: commit writes sections in hash order, sorts keys and drops trailing comments. Indented comments have been kept since the fix for 3612465. A prototype that preserves file order and trailing comments is ready.
  • Side note (not part of the patch): ini::revert passes the handle name instead of the channel to _setfileenc (it fails when -encoding was used). namespace eval ... {array set data {}} does not clear the arrays, so revert does not actually discard changes.
⬜2907556fff2009importFile fails, if the client filename contains a ;patchREPRODUCED (9.1, 8.6). parseMimeValue {form-data; name="f"; filename="a;b.txt"} gives filename "a, and importFile -client returns "a.

Open ticket 2907556fff · Patch · ncgi · filed 2009-12-02

Full verdict

REPRODUCED (9.1, 8.6). parseMimeValue {form-data; name="f"; filename="a;b.txt"} gives filename "a, and importFile -client returns "a.

  • Module ncgi 1.4.6
  • Tested parseMimeValue and a full multipart parse with importFile, on 9.1 and 8.6
  • Cause ncgi.tcl:parseMimeValue splits the header on every ";" before looking at quotes
  • Prototype prototypes/2907556fff.patch (78 lines, about 49 changed. The attached regexp approach was reworked to keep the backslash-quote handling from 063e7a95. New tests ncgi-13.9, 13.10, 16.7 fail without the fix. Existing 13.x tests unchanged.)
  • Suggested action apply prototype
  • Suggested ticket comment parseMimeValue split on every ";" even inside quotes. It now matches quoted values as a whole, based on the attached patch.
🔧2853152fff2009fix async dns behaviourpatchREPRODUCED (9.1, 8.6), by code and test. Finish closes state(sock) but keeps the name, and a later dns::cleanup closes it again. The attached…

Open ticket 2853152fff · Patch · dns · filed 2009-09-06

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 822c02055e on dns-async-double-close, not merged: Fix 2853152fff: an async dns query closed its socket twice. Clear state(sock) after closing, and skip an empty socket in cleanup.
  • 💬 2026-11-16 serhiy.storchaka: The 2009 patch was never applied, and the double close is still there. Proposed fix in branch dns-async-double-close, following the patch: clear state(sock) after closing, and skip an empty socket in cleanup. Test dns-5.1.
Full verdict

REPRODUCED (9.1, 8.6), by code and test. Finish closes state(sock) but keeps the name, and a later dns::cleanup closes it again. The attached patch was never applied. A real name reuse could not be forced here (Unix socket names are pointers), so the test watches the close calls.

  • Module dns 1.6.1
  • Tested local TCP DNS server on 127.0.0.1, random port; dns tests on 9.1 and 8.6
  • Cause dns.tcl ::dns::Finish and ::dns::cleanup.
  • Prototype prototypes/2853152fff.patch (53 lines, test dns-5.1). Clears state(sock) after closing, and cleanup skips an empty socket.
  • Suggested action apply prototype
  • Suggested ticket comment The 2009 patch was never applied and the double close is still there. Re-created against trunk, with a test.
🔧2848168fff2009signmask incorrect in ::crc::Crc and ::crc::Crc32_tclpatchREPRODUCED on Windows (9.1, 8.6): tcl_platform(wordSize) is 4, ::crc::signbit is 2147483648, and CRC-32/BZIP2 of "123456789" through `::crc::Cr…

Open ticket 2848168fff · Bug · crc · filed 2009-09-01

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 69f713b9da on crc-32bit-wordsize, not merged: Fix 2848168fff: ::crc::Crc gave wrong results for non-reflected 32-bit CRCs where tcl_platform(wordSize) is 4 (Windows): the signbit mask cleared bit 24. Valu
  • 💬 2026-11-15 serhiy.storchaka: Confirmed on Windows with Tcl 9.1 and 8.6, where tcl_platform(wordSize) is 4: CRC-32/BZIP2 of "123456789" gives F9E2E667 instead of FC891918. The values are always masked non-negative, so the sign mask is not needed. Proposed fix in branch crc-32bit-wordsize, which removes it; test crc16-10.0.
Full verdict

REPRODUCED on Windows (9.1, 8.6): tcl_platform(wordSize) is 4, ::crc::signbit is 2147483648, and CRC-32/BZIP2 of "123456789" through ::crc::Crc gives F9E2E667 instead of FC891918.

  • Module crc16 1.1.5, crc32 1.3.4
  • Tested tclsh9.1, 8.6, 8.5 on Linux; Windows emulated by presetting ::crc::signbit 2147483648
  • Cause crc16.tcl ::crc::Crc and crc32.tcl Crc32_tcl mask with ~signbit>>7, which clears bit 24 on Windows. Since Tcl 8.5 has bignums and the values are always masked non-negative, the mask is not needed at all.
  • Prototype prototypes/2848168fff.patch (30 lines, tests crc16-10.0/10.1). Removes the signmask.
  • Suggested action apply prototype
  • Suggested ticket comment With Tcl 8.5+ integers the sign mask is not needed. It only breaks user-defined non-reflected 32-bit CRCs via ::crc::Crc where wordSize is 4 (Windows). The built-in commands were not affected. Patch removes it.
Windows VM, run 1
  • Verdict REPRODUCED on Windows (9.1, 8.6): tcl_platform(wordSize) is 4, ::crc::signbit is 2147483648, and CRC-32/BZIP2 of "123456789" through ::crc::Crc gives F9E2E667 instead of FC891918.
  • Tested ::crc::Crc_table 32 0x04C11DB7 0 + ::crc::Crc 123456789 32 tbl 0xFFFFFFFF 0xFFFFFFFF 0 with tclsh91 and tclsh86t. The prototype prototypes/2848168fff.patch removes the mask; its test crc16-10.0 emulates this signbit on Linux.
⬜2526629fff2009SPF package is apparently dysfunctionalpatchREPRODUCED (9.1, 8.6)

Open ticket 2526629fff · Bug · spf · filed 2009-01-21

Full verdict

REPRODUCED (9.1, 8.6)

  • Module spf 1.1.2 (modules/dns/spf.tcl)
  • Tested local DNS-over-TCP server on 127.0.0.1 (random port) serving two TXT records, with dns::configure -protocol tcp. spf::SPF returns oogle-site-verification=abc=spf1 ip4:... and spf::spf returns none for every address.
  • Cause spf.tcl ::spf::SPF strips the first character of each rdata (dns.tcl ReadString already removes the length byte) and concatenates all TXT records, including non-SPF ones. Its error path also uses undefined $log/$r.
  • Prototype prototypes/2526629fff.patch (113 lines, tests spf-7.1..7.4). It keeps the whole rdata, selects only v=spf1 records (RFC 7208 4.5), returns none if there is none, and raises a permanent error if there are several. It also fixes the error-path variables. spf.test passes on 9.1/8.6; the new tests fail without the fix. Verified end-to-end with the local DNS server: + / - results.
  • Suggested action apply prototype
  • Suggested ticket comment Still broken: the first rdata char is dropped and non-SPF TXT records are concatenated in. Proposed fix (keep rdata whole, select the v=spf1 record) with tests spf-7.x is ready.
🔧2526633fff2009SPF package fails with hosts with IPv6 specificationspatchREPRODUCED (9.1, 8.6)

Open ticket 2526633fff · Bug · spf · filed 2009-01-21

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 0389fb7aa9 on spf-ipv4-ipv6-mismatch, not merged: Fix 2526633fff: SPF checks raised an error when an IPv4 address was compared with an IPv6 one. Such addresses now simply do not match.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch spf-ipv4-ipv6-mismatch: addresses of the other family no longer match, instead of raising an error. Tests spf-4.6 to 4.9.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module spf 1.1.2
  • Tested spf::Spf 2001:db8::1 ... "v=spf1 ip4:... ip6:... -all" gives error in "ip4": type mismatch. Same for an IPv4 sender with ip6 first, and for a/mx with an IPv6 sender.
  • Cause spf.tcl _ip4, _ip6, _a, _mx and _ptr call ip::equal on addresses of different families, which throws.
  • Prototype prototypes/2526633fff.patch (132 lines, tests spf-4.6..4.9). A new helper IpMatch returns 0 when the families differ, and _ptr gets the same check. spf.test passes on 9.1/8.6; the new tests fail without the fix. The patch is independent of 2526629fff; both apply together (47/47 pass).
  • Suggested action apply prototype. Note: a/mx still only look up A records (no AAAA), a separate limitation.
  • Suggested ticket comment Confirmed. Proposed fix: mechanisms skip addresses of the other family instead of erroring. Tests spf-4.6..4.9.
⬜2440373fff2008broken compatibility for cookie with Safari + OS X.5.6patchREPRODUCED (9.1, 8.6). setCookie still emits SID=foo ; path=/ ; ... secure , with spaces before ";" and a trailing space. That is not RFC 6265 sy…

Open ticket 2440373fff · Bug · ncgi · filed 2008-12-17

Full verdict

REPRODUCED (9.1, 8.6). setCookie still emits SID=foo ; path=/ ; ... secure , with spaces before ";" and a trailing space. That is not RFC 6265 syntax (name=value; attr=val). Modern browsers trim the value, but the output is still malformed.

  • Module ncgi 1.4.6
  • Tested ncgi::setCookie ...; set ncgi::cookieOutput and the existing ncgi-11.2/12.4 header tests, on 9.1 and 8.6
  • Cause ncgi.tcl:setCookie string formatting
  • Prototype prototypes/2440373fff.patch (76 lines, about 24 changed. New test ncgi-12.5. Updates the expected output of ncgi-11.2/12.4. All three fail without the fix.)
  • Suggested action apply prototype
  • Suggested ticket comment setCookie still emitted "name=value ;". It now produces RFC 6265 syntax: "name=value; path=...; domain=...; expires=...; secure".
🔧2426988fff2008linking to array variable problempatchREPRODUCED (9.1, 8.6)

Open ticket 2426988fff · Bug · struct :: matrix · filed 2008-12-14

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 09d3df39b0 on matrix-link-stale-elements, not merged: Fix 2426988fff: an array linked to a struct::matrix kept stale elements after deleting rows or columns or transposing. Pass unsets on to the linked array.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch matrix-link-stale-elements: the matrix data is now traced for unsets too, so cells removed by delete row/column or transpose are also removed from the linked arrays. Tests matrix-9.15 and matrix-9.16.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module struct::matrix 2.2
  • Tested a 2x3 matrix linked to a and, with -transpose, to b, then delete row 1. The linked arrays keep stale elements for the old last row (0,2, 1,2). The same happens after delete column, delete rows/columns n and transpose of a non-square matrix.
  • Cause matrix.tcl _link puts only a write trace on the internal data array. Moving cells down is propagated, but unsetting the vacated last row/column is not. Fix: trace {write unset} and make MatTraceOut unset the array element on an element unset. A whole-array unset is ignored, and the existing cross-array unset propagation is unchanged.
  • Prototype prototypes/2426988fff.patch (112 lines, tests matrix-9.15, matrix-9.16; matrix-9.14 updated for the new trace info). New tests fail without the fix. The struct module passes on 9.1 and 8.6.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: the linked array is traced only for writes, so cells removed by delete row/column or transpose stay in the linked array. Proposed fix: also trace unsets and propagate them to the linked arrays (patch with tests available).
⬜2144802fff2008snit's from command underdocumented, and possible flawpatchREPRODUCED (9.1, 8.6; snit 1 and 2)

Open ticket 2144802fff · Bug · snit · filed 2008-10-03

Full verdict

REPRODUCED (9.1, 8.6; snit 1 and 2)

  • Module snit 2.3.4 / 1.4.3 (main2.tcl / main1.tcl ::snit::RT.from)
  • Tested dog create spot -foo -bar -snarf 0 -foo {a b} with [from args -bar xyzzy] returns -snarf and leaves {-foo 0 -foo {a b}}; tclsh9.1, tclsh8.6, snit.tcl + snit2.tcl.
  • Cause main1.tcl/main2.tcl RT.from uses lsearch -exact on the whole list, so it matches values at odd positions too, and it only removes the first occurrence.
  • Prototype prototypes/2144802fff.patch (175 lines, test from-1.3; snit.man documents the rules). Implements what Will Duquette proposed in 2008: look only at even positions, remove all occurrences, return the last value. snit.test: 0 failures on 9.1/8.6; from-1.3 fails without the fix.
  • Suggested action apply prototype. Small compatibility risk: code that calls from on lists with a positional prefix of odd length would stop finding the option.
  • Suggested ticket comment Still present. Proposed fix (only even positions are option names, all occurrences removed, last value wins, as Will suggested) plus doc update and test from-1.3 is ready.
⬜2113728fff2008"$type info instances" corrupted after object-creation errorpatchREPRODUCED (9.1, 8.6; Xvfb)

Open ticket 2113728fff · Bug · snit · filed 2008-09-16

Full verdict

REPRODUCED (9.1, 8.6; Xvfb)

  • Module snit 2.3.4 / 1.4.3
  • Tested myWidget .x twice, then myWidget info instances gives can't read "instance": no such variable. Also when the name collides with a plain frame. snit 1 and 2.
  • Cause main1.tcl/main2.tcl: RT.widget.typemethod.create. The instance namespace is created before installhull, which runs outside the catch that cleans up failed constructions. A failed hull creation leaves a Snit_inst* namespace without Snit_instance.
  • Prototype prototypes/2113728fff.patch (87 lines, test widget-2.2). Catches the installhull failure, deletes the half-made namespace, and re-throws. snit.test under wish on Xvfb: 0 failures on 9.1/8.6 (snit 2). snit 1 counts unchanged. The new test fails without the fix.
  • Suggested action apply prototype
  • Suggested ticket comment Still present. A failed hull creation leaves a half-made instance namespace behind. Proposed fix with test in prototypes/2113728fff.patch.
⬜2063144fff2008Snit method name collisionpatchREPRODUCED (9.1, 8.6)

Open ticket 2063144fff · Bug · snit · filed 2008-08-20

Full verdict

REPRODUCED (9.1, 8.6)

  • Module snit 2.3.4 / 1.4.3
  • Tested typemethod {a_b c} / {a b_c} and the same for methods. The second overwrites the first (both return 2). snit 1 and 2.
  • Cause main1.tcl/main2.tcl: the proc names of hierarchical methods use [join $method _].
  • Prototype prototypes/2063144fff.patch (117 changed lines, mechanical, same edits in main1 and main2; tests htypemethod-1.4a, hmethod-1.4a). New RT.MangleMethod escapes %→%25 and _→%5f inside tokens before joining with _. Names without _/% keep the same proc names. Adds an internal %J substitution, so the documented %j (underscore join in delegate patterns) is unchanged. It covers compile-time and runtime snit::method/snit::typemethod, dispatch, and info args/body/default. snit.test (snit 2): 0 failures on 9.1/8.6. snit 1 failure counts unchanged and the new tests pass. The new tests fail without the fix. Larger than the usual limit, but mechanical.
  • Suggested action apply prototype (review the %J naming)
  • Suggested ticket comment Still present. Proposed fix: escape _ and % inside tokens in the generated proc names; documented %j unchanged (prototypes/2063144fff.patch, with tests).
🔧2055837fff2008Bad InteropIFD directory causes crashpatchREPRODUCED (9.1, 8.6)

Open ticket 2055837fff · Bug · jpeg · filed 2008-08-17

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 8f6cc10bd4 on jpeg-bad-exif-offsets, not merged: Fix 2055837fff: jpeg::getExif raised an error for Exif entries whose offsets point past the end of the data, e.g. a bad InteropIFD. Skip such entries.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch jpeg-bad-exif-offsets: Exif entries whose offsets point outside the data are skipped instead of raising an error. Tests jpeg-43.0 and jpeg-43.1.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module jpeg 0.7 (modules/jpeg/jpeg.tcl)
  • Tested Generated minimal JPEGs with an Exif value offset past the end, and with an InteropIFD (a005) pointer past the end. getExif fails with expected integer but got "" or can't read "num": no such variable.
  • Cause jpeg.tcl:_exif reads values and sub-IFDs at file offsets without checking that the read succeeded and returned enough bytes.
  • Prototype prototypes/2055837fff.patch (35 lines, tests jpeg-43.0, jpeg-43.1). Returns an empty block if the entry count cannot be read, stops on a short entry, and skips values not fully inside the data. Tests run with TCLLIBPATH=modules (jpeg.test needs fileutil/cmdline).
  • Suggested action apply prototype (a self-referencing sub-IFD loop would still recurse, which could be guarded separately).
  • Suggested ticket comment Still reproducible with an Exif offset that points past the end of the data. A prototype that skips such entries instead of throwing is ready.
⬜1968777fff2008Locale problemspatchREPRODUCED (9.1, 8.6, 8.5)

Open ticket 1968777fff · Bug · doctools · filed 2008-05-21

Full verdict

REPRODUCED (9.1, 8.6, 8.5)

  • Module doctools 1.6.1 (checker.tcl, checker_idx.tcl, checker_toc.tcl); the same issue exists in doctools2 (doctools::msgcat)
  • Tested formatted a manpage with an error under LANG=pl_PL.UTF-8 on 9.1/8.6/8.5. The output is unknown error code "deflist" (for locale pl_pl). LANG=C and de_DE give real messages. doctools2 has the same problem: doctools::idx::parse text foo under pl_PL gives error on line 1.3: docidx/plaintext instead of the message.
  • Cause the checkers mcload the mpformats/*.msg catalogs, but those exist only for c/en/de/fr. There is no ROOT catalog, so other locales find nothing, and the overridden mcunknown prints the code.
  • Prototype prototypes/1968777fff.patch (+64 lines, mostly 3x the same 12 lines; tests doctools-9.0.1, docidx-9.0.1, doctoc-9.0.1). After mcload, each checker reads c.msg and defines its messages for the root locale {}. msgcat then falls back to C for unknown locales or incomplete catalogs. Changing mclocale inside mcunknown does not work: msgcat 1.6+ reloads package locales and calls file, which is a formatter command in the checker interp. The new tests fail on 9.1 and 8.6 without the fix and pass with it. The doctools module is 602/602 on both. The doctools2 (doctools::msgcat::init) side is not covered; fixing it there needs a root-locale fallback in doctools2base/msgcat.tcl.
  • Side finding doctools2base doctools-msgcat-2.0 fails when the user's LANG is not C (here uk_UA). The test sets env(LANG) after msgcat was already loaded. This is a test-environment problem.
  • Suggested action apply the prototype, and decide whether to do the same for doctools2
  • Suggested ticket comment Still reproducible with any locale that has no catalog (e.g. pl_PL). There is no root-locale catalog, so the error code is printed. Proposed fix: load the C messages as the root catalog in the checkers. doctools2's msgcat has the same issue.
🔧1924065fff2008tiff - addEntry doesn't overwritepatchREPRODUCED (9.1, 8.6)

Open ticket 1924065fff · Bug · tiff · filed 2008-03-24

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in b15c6607e4 on tiff-addentry-overwrite, not merged: Fix 1924065fff: tiff::addEntry did not replace an existing entry with the same tag, and nametotag matched only part of the name. Remove the knownBug constrain
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch tiff-addentry-overwrite. addEntry stored the entry under the whole entry list instead of its tag, so the file got a second entry with the same tag; nametotag also checked the whole entry instead of the tag. The tiff-23 tests are enabled, and tiff-23a tests numeric tags.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module tiff 0.2.3
  • Tested tiff.test with -constraints knownBug: tiff-23.* return {Artist Andreas} {Artist Andreas}. Run directly, tiff.test needs TCLLIBPATH, because its support block lacks cmdline for fileutil. sak works.
  • Cause tiff.tcl:addEntry stores the new entry as ifd($e), where $e is the whole {tag format value} list, not the tag. The new entry therefore lands under a separate key, the file gets two entries with the same tag, and getEntry returns the old one. Also, nametotag checks the hex-tag pattern against $x (the whole entry) instead of $y, so addEntry $f {{013b 2 AK}} fails with "unknown tag".
  • Prototype prototypes/1924065fff.patch (71 lines). Uses ifd([lindex $e 0]), fixes the nametotag pattern, removes knownBug from tiff-23.* and adds tiff-23a.* (overwrite by numeric tag). Without the fix 6 tests fail. With it, 285 tests, 0 failures on 9.1 and 8.6. Unrelated, not fixed: _copyData's EXIF branch uses a stale $x in lindex $ifd($x) 0 and should use ifd(8769).
  • Suggested action apply prototype
  • Suggested ticket comment The cause is that addEntry stored the new entry under the whole entry list instead of its tag, so the file got a second entry with the same tag. A fix is prepared, and it enables the tiff-23 tests.
⬜1886636fff2008[clock] inside snit::widgetadaptorpatchREPRODUCED (9.1, 8.6)

Open ticket 1886636fff · Bug · snit · filed 2008-02-04

Full verdict

REPRODUCED (9.1, 8.6)

  • Module snit 2.3.4 / 1.4.3
  • Tested option -foo[clock format 0 ...] in a type body. 9.1: unknown subcommand "getsystemlocale". 8.6: can't read "TclLibDir". Both snit 1 and 2.
  • Cause main1.tcl/main2.tcl: Comp.Init. The compiler slave interp renames proc/variable, which breaks the Tcl-coded parts of clock/msgcat.
  • Prototype prototypes/1886636fff.patch (68 lines, test option-1.9a). Adds $compiler alias clock ::clock in both snit versions, as kennykb suggested. snit.test: 0 failures on 9.1 and 8.6 (snit 2). snit 1 runs (copy of snit.test switched to snit 1) have the same pre-existing failures as before (27 on 9.1, 3 on 8.6). The new test fails without the fix. clock was already a reserved word, so macros are unaffected.
  • Suggested action apply prototype
  • Suggested ticket comment Still fails on 8.6 and 9.x. Proposed fix: alias clock from the main interp into the compiler (prototypes/1886636fff.patch, with test).
⬜1871959fff2008::ncgi::query breaks encodingREPRODUCED (9.1, 8.6) for multipart/form-data text fields: a field sent as UTF-8 "йq" comes back from ncgi::value as the 3-character byte string …

Open ticket 1871959fff · Bug · ncgi · filed 2008-01-15

Full verdict

REPRODUCED (9.1, 8.6) for multipart/form-data text fields: a field sent as UTF-8 "йq" comes back from ncgi::value as the 3-character byte string "йq". Percent-encoded urlencoded data is fine (decode handles UTF-8).

  • Module ncgi 1.4.6
  • Tested real CGI invocation with REQUEST_METHOD=POST, CONTENT_LENGTH and a body piped into stdin, on 9.1 and 8.6. The test case zip linked in the ticket is a dead link.
  • Cause ncgi.tcl:query reads stdin with -translation binary. That is required, because CONTENT_LENGTH counts bytes and file uploads must stay binary. But ncgi::multipart never converts non-file parts with the part's charset (or UTF-8).
  • Prototype none (needs a design decision: decoding text parts would change values for applications that already do encoding convertfrom utf-8 themselves; possible options are an opt-in flag or decoding only parts without a filename)
  • Suggested action keep open; human decision on API
  • Suggested ticket comment Still true for multipart/form-data: text fields are returned as raw UTF-8 bytes, while file parts must stay binary. Fixing it needs either an opt-in or a compatibility decision.
⬜1799702fff2007::ip::reduceToAggregates with weird netmask brokenpatchREPRODUCED (9.1, 8.6)

Open ticket 1799702fff · Bug · ip · filed 2007-09-21

Full verdict

REPRODUCED (9.1, 8.6)

  • Module ip 1.5.1 (modules/dns/ipMore.tcl)
  • Tested ip::reduceToAggregates {1.2.0.4/255.255.0.255 1.2.3.0/255.255.255.0} gives 1.2.0.4/16 on 9.1 and 8.6. Also found: a single-element list fails ("cannot use non-numeric string "" as left operand of "&"").
  • Cause ipMore.tcl:reduceToAggregates drops any overlapping prefix with the numerically larger mask. That only equals containment for contiguous masks. nativeToPrefix then prints 0xffff00ff as "/16". The single prefix case breaks because prefixToNative returns a non-listified result for one element.
  • Prototype prototypes/1799702fff.patch (83 lines, tests ::ip:reduceToAggregates-1799702, -1799702-2). A prefix is dropped only if another prefix contains it (mask subset and same network), keeping the first of equal ones. Non-contiguous masks are output in dotted form. A single prefix is wrapped. Existing ip/ipMore tests pass on 9.1 and 8.6.
  • Suggested action apply prototype. Note that ip::collapse also turns the same input into 1.2.0.0/16. A non-contiguous mask cannot be expressed there; consider rejecting such masks.
  • Suggested ticket comment Still reproducible. reduceToAggregates compares mask values instead of checking containment. A prototype that only removes prefixes fully covered by another one is ready.
⬜1777399fff2007ftp cause memory leakpatchREPRODUCED (9.1, 8.6), partly: the leak is real, but only for failed ftp::Open

Open ticket 1777399fff · Bug · ftp · filed 2007-08-20

Full verdict

REPRODUCED (9.1, 8.6), partly: the leak is real, but only for failed ftp::Open

  • Module ftp 2.4.14
  • Tested 600 Open/Get(file)/Get(-variable)/Close cycles against a local ftpd, passive and active, on 9.1 and 8.6. No growth: the ftp vars, channels and RSS stay flat. A failed ftp::Open (connection refused, or the server closes before the greeting) returns -1 but leaves the ::ftp::ftpN state array behind, one per attempt. The reporter's script retries Open in a loop and never cleans up after a failure, so it leaks exactly this way.
  • Cause ftp.tcl ::ftp::Open. When OpenControlConn fails, only ftp(State) is unset. When the login fails, Close is called, but it returns early ("Not connected!") because ftp(State) is already gone. Neither path unsets the array or closes the control socket.
  • Prototype prototypes/1777399fff.patch (ftp.tcl 9 changed lines; tests ftp-3.0, ftp-3.1)
  • Suggested action apply prototype, close as Fixed
  • Suggested ticket comment No leak in normal Open/Get/Close cycles today. But every failed ftp::Open (connection refused or dropped) left its session array behind, which matches the retry loop in the attached script. Proposed fix with tests in the prototype.
⬜1763707fff2007::ftp::Get returns "" in mode inline + errorpatchREPRODUCED (9.1, 8.6)

Open ticket 1763707fff · Bug · ftp · filed 2007-07-30

Full verdict

REPRODUCED (9.1, 8.6)

  • Module ftp 2.4.14
  • Tested ftp::Get $h nosuchfile -variable v against a local tcllib ftpd, in passive and active mode. It returns "", but the manual says 0. Also found: in active mode, against a server that opens the data connection right after PORT (ftpd, pyftpdlib), an error reply to RETR (e.g. 550) hangs ftp::Get forever. The control timeout fires inside a nested vwait on state.data and cannot release it.
  • Cause ftp.tcl ::ftp::Get has an inline special case that returns "" and skips CloseDataConn. In the get_sent/reget_sent error branches, state.data is unset only in passive mode, so WaitComplete waits for a data connection that the server never closes. In active mode the data socket was also never recorded, so it could not be closed.
  • Prototype prototypes/1763707fff.patch (ftp.tcl ~20 changed lines; tests ftp-1.1-{passive,active} (return value) and ftp-1.2-{passive,active} (no hang on 550, with a watchdog))
  • Suggested action apply prototype, close as Fixed
  • Suggested ticket comment Still present: on error Get -variable returned "" instead of 0 and did not close the data connection. In active mode a refused RETR could even hang. Proposed fix with tests in the attached prototype (branch to be created).
🔧1749395fff2007Patch to ignore directories in Mail BoxespatchREPRODUCED (9.1, 8.6)

Open ticket 1749395fff · Patch · pop3d · filed 2007-07-06

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 9e88cd99b3 on pop3d-dbox-skip-dirs, not merged: Fix 1749395fff: pop3d::dbox counted subdirectories of a mailbox, e.g. CVS, as messages. Glob only files.
  • 💬 2026-11-16 serhiy.storchaka: Proposed fix in branch pop3d-dbox-skip-dirs: the idea of the patch, now with glob -types f -directory, which is fine since tcllib needs Tcl 8.5. Test pop3-dbox-8.4.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module pop3d::dbox 1.0.3 (modules/pop3d/pop3d_dbox.tcl)
  • Tested a mailbox with a CVS/ subdirectory. STAT reports 2 messages (the directory counts, with its 4096-byte size), and RETR of that "message" fails.
  • Cause pop3d_dbox.tcl:_lock builds the message list with glob -nocomplain [file join $dir *]. AK's 2007 objection (Tcl 8.3 lacks -types/-directory) no longer applies, since tcllib needs 8.5+.
  • Prototype prototypes/1749395fff.patch (51 lines, test pop3-dbox-8.4). It uses glob -nocomplain -types f -directory $dir *. The test fails without the fix, and the pop3d suite passes on 9.1 and 8.6.
  • Suggested action apply prototype
  • Suggested ticket comment Still present. The Tcl 8.3 concern no longer applies (tcllib needs 8.5), so the proposed fix uses glob -types f -directory; it has a test.
⬜1747945fff2007Patch to avoid inflation of the 'statTime' arraypatchREPRODUCED (9.1, 8.6): statTime($name) still grows by one element per call, so memory grows without bound in long runs.

Open ticket 1747945fff · Patch · profiler · filed 2007-07-04

Full verdict

REPRODUCED (9.1, 8.6): statTime($name) still grows by one element per call, so memory grows without bound in long runs.

  • Module profiler 0.7
  • Tested profiled foo called 100 times. The total list length in the profiler arrays grows by 100.
  • Cause profiler.tcl:leaveHandler does lappend ::profiler::statTime($name) $t, and stats recomputes from all samples at print/dump time.
  • Prototype prototypes/1747945fff.patch (219 lines, about 80 changed). It keeps only statSumSq($name) and adds StatsFromSums N sum sumsq, which reproduces the existing stats result exactly, including the integer mean, using sum((v-m)^2) = S - 2mT + Nm^2. The public stats {args} proc is unchanged. Tests: profiler-12.0 (state does not grow) and profiler-12.1 (same result as stats). Both fail without the fix; the suite passes on 9.1 and 8.6. The attached 2007 patch changes the numbers (float mean) and the stats signature, so it was not used. The open branch pbrooks-2026-11-windows-profiler-fixes only touches tZero/tMark, so there is no overlap.
  • Suggested action apply prototype
  • Suggested ticket comment Still present: statTime keeps every sample. Proposed fix keeps only the sum of squares (totalRuntime already has the sum) and gives identical mean/stddev/cov output; it has tests.
⬜1743868fff2007Fix extra line between message headerspatchREPRODUCED (9.1, 8.6)

Open ticket 1743868fff · Patch · pop3d · filed 2007-06-26

Full verdict

REPRODUCED (9.1, 8.6)

  • Module pop3d 1.2.0 (modules/pop3d/pop3d.tcl)
  • Tested local pop3d (port 0) + pop3d::dbox + raw binary socket client; RETR and TOP output on 9.1 and 8.6. Every message line arrives as CR CR LF. Because of that, TOP n k never finds the header/body separator and sends the whole message.
  • Cause pop3d.tcl:Transfer. mime::buildmessage (mime 1.7.2, memchan with -translation crlf) already returns CRLF lines, and pop3d writes them to a socket configured with -translation crlf. The TOP branch splits on \n and looks for an empty line, but the separator is \r. Two more TOP problems turn up once the separator is found: incr limit -1 makes TOP n k send only k-1 body lines, and H_top treats TOP n 0 as "whole message" while RFC 1939 says headers only.
  • Prototype prototypes/1743868fff.patch (135 lines; tests pop3-srv-8.0.0..2, a self-contained raw-socket client). It normalizes the buildmessage output to LF in both branches, removes the off-by-one and makes TOP n 0 return headers only. The new tests fail without the fix (3 failures) and the pop3d suite passes on 9.1 and 8.6 (155/155). The attached 2007 patch (reversed diff) instead switched the socket to lf and added \r by hand everywhere. The prototype is less invasive.
  • Suggested action apply prototype
  • Suggested ticket comment Still present: mime::buildmessage returns CRLF and pop3d writes it to a crlf-translated socket, so clients get CR CR LF. This also breaks TOP, which sends the full message. Proposed fix normalizes to LF before sending and fixes the TOP line count; it has a test.
⬜1656200fff2007struct man name inconsistenciesREPRODUCED (still inconsistent)

Open ticket 1656200fff · Bug · struct · filed 2007-02-09

Full verdict

REPRODUCED (still inconsistent)

  • Module struct (doc files)
  • Tested manpages are generated from file names. Some have a prefix (struct_list, struct_map, struct_set, struct_tree, struct_tree1), others not (disjointset, graph, graph1, graphops, matrix, pool, prioqueue, queue, record, skiplist, stack). See idoc/man/files/modules/struct/*.n.
  • Cause historical; Andreas proposed in 2007 to prefix all of them.
  • Prototype none (renaming .man files affects docs, links and installed names; a maintainer decision)
  • Suggested action human decision: rename all to struct_*.man, or close as Won't Fix
  • Suggested ticket comment Still inconsistent (11 pages unprefixed, 5 with struct_). Renaming all to struct_* as proposed in 2007 is easy but changes the installed man page names.
⬜1606391fff2006parse gets confused inside script tagspatchREPRODUCED (9.1, 8.6)

Open ticket 1606391fff · Bug · htmlparse · filed 2006-11-30

Full verdict

REPRODUCED (9.1, 8.6)

  • Module htmlparse 1.2.3
  • Tested <script>...i<window.document...</script> on 9.1 and 8.6: the parser reports a bogus tag window.document.forms.length;.
  • Cause htmlparse.tcl parse finds tags with one regsub over the whole document. The raw text of <script>/<style> is not treated specially.
  • Prototype prototypes/1606391fff.patch (68 lines; PrepareHtml converts </> inside script/style to &lt;/&gt;, the same way {, } and \ are already turned into entities in all text; test htmlparse-12.0). The test fails without the fix. The 59 tests pass on 9.1 and 8.6. Limitation: in incremental (-incvar) mode, a script element split across chunks is not handled.
  • Suggested action apply prototype
  • Suggested ticket comment Still present. Proposed fix: the contents of script and style elements are protected from tag recognition, and </> in them are delivered as entities, like the existing brace escapes.
⬜1519755fff2006Unable to delete recordpatchREPRODUCED (9.1, 8.6)

Open ticket 1519755fff · Bug · struct :: record · filed 2006-07-10

Full verdict

REPRODUCED (9.1, 8.6)

  • Module struct::record 1.2.4
  • Tested the attached itcl script reduced to plain Tcl. The reporter calls record delete instance $obj.sub1.reserved1, which names a field, not an instance. It silently does nothing. record delete instance junk and record delete record junk also return without error. struct_record.man says both "will return an error if the ... does not exist".
  • Cause record.tcl Delete has #error "$item_ is not a instance" commented out, and the record branch has no existence check. Fields cannot be deleted at all, by design. Separately, deleting a nested instance ($obj.sub1) directly leaves the parent broken (show values then fails); the prototype does not change that.
  • Prototype prototypes/1519755fff.patch (55 lines, tests record-9.0-tkt-1519755fff, record-9.1-tkt-1519755fff). New tests fail without the fix. The struct module passes on 9.1 and 8.6. No other tcllib code calls record delete.
  • Suggested action apply prototype (small behaviour change: deleting a missing instance or record now raises an error, as documented)
  • Suggested ticket comment The script deletes a field ($objTest.sub1.reserved1), not an instance. Record fields cannot be deleted, but the call should have raised an error, as documented, instead of silently doing nothing. Proposed fix: raise an error for unknown instances and records (patch with tests).
⬜1508656fff2006default fsCmd fails to store.patchREPRODUCED (9.1 with the 34a589f660 fix, 8.6)

Open ticket 1508656fff · Bug · ftpd · filed 2006-06-19

Full verdict

REPRODUCED (9.1 with the 34a589f660 fix, 8.6)

  • Module ftpd 1.4.1
  • Tested a default-configured ftpd with a world-writable docroot. STOR newfile.txt gives "550 Permission denied". MKD newdir and RMD dir are always denied too.
  • Cause ftpd.tcl ::ftpd::fileAuth. Fs exists is false for directories, so the "check the parent directory" branch for new paths never runs, and the final Fs exists $path check denies everything that is not an existing file.
  • Prototype prototypes/1508656fff.patch (on top of 34a589f660.patch; fileAuth rewritten, ~25 lines; tests ftpd-2.0, ftpd-2.1 (unix constraint)). Checked end to end: STOR/MKD on a new name work afterwards.
  • Suggested action apply prototype (after 34a589f660), close as Fixed
  • Suggested ticket comment Still true: with the default authFileCmd, creating files and directories (and RMD) was always denied, because fsCmd exists is false for directories. Proposed fix with tests in the prototype.
🔧1491894fff2006comm "-listen 0" causes race conditionsREPRODUCED (9.1, 8.6)

Open ticket 1491894fff · Bug · comm · filed 2006-05-19

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in 09fc83dd5b on comm-lost-conn-peer, not merged: Fix 4811399edc, 1491894fff: comm raised an error when a client disconnected after the "commIncoming race condition" warning. commLostConn now clears the pen
  • 💬 2026-11-16 serhiy.storchaka: The same bug as 4811399edc; proposed fix in branch comm-lost-conn-peer.
Full verdict

REPRODUCED (9.1, 8.6)

  • Module comm 4.7.3
  • Tested 9.1, 8.6: two non-listening channels in one process sending to a server: server prints "commIncoming race condition: 0"; when they disconnect, bgerror "can't read comm(...,pending,0)" (same as 4811399edc)
  • Cause comm.tcl commIncomingOffered/commNewConn: all non-listening peers on a host get id 0; the second connection is reported as a race, and commLostConn treats every fid as the owner of peers/pending for the id
  • Prototype prototypes/4811399edc.patch (shared; suppresses the warning for remote id 0 and fixes commLostConn; test comm-4811399edc). Distinguishing such senders is by design not possible (id 0 means "cannot be called back"); replies still go on the right socket
  • Suggested action apply prototype, close together with 4811399edc
  • Suggested ticket comment Non-listening peers all share id 0, so they cannot be told apart by id. The bogus race warning and the error on disconnect are fixed in the patch for 4811399edc.
⬜1123327fff2005histograms do not handle negative reals correctly.patchREPRODUCED (9.1, 8.6): counting 9.1 and -9.2 into -hist 10 gives 0 2, expected -1 1 0 1. Negative integers are bucketed correctly, because in…

Open ticket 1123327fff · Bug · counter · filed 2005-02-15

Full verdict

REPRODUCED (9.1, 8.6): counting 9.1 and -9.2 into -hist 10 gives 0 2, expected -1 1 0 1. Negative integers are bucketed correctly, because integer division already floors.

  • Module counter 2.0.6
  • Tested tclsh9.1, tclsh8.6; counter.test
  • Cause counter.tcl ::counter::count, branch -hist: int($delta / bucketsize) truncates toward zero for reals. The ticket's change to the HTML labels is not needed: with floor, label i*bucketsize is the correct lower bound of bucket i.
  • Prototype prototypes/1123327fff.patch (11 lines, test counter-hist-1.2). Uses int(floor(...)).
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed on trunk: reals between -bucketsize and 0 fall into bucket 0. Fixed by flooring the quotient. The HTML label change is not needed.
⬜1011102fff2004counter - dependent on ncgi & html, not in code, nor docspatchREPRODUCED (9.1, 8.6): counter::histHtmlDisplay fails with invalid command name "ncgi::value" unless the caller has already loaded ncgi and html.

Open ticket 1011102fff · Bug · counter · filed 2004-08-18

Full verdict

REPRODUCED (9.1, 8.6): counter::histHtmlDisplay fails with invalid command name "ncgi::value" unless the caller has already loaded ncgi and html.

  • Module counter 2.0.6
  • Tested tclsh9.1, tclsh8.6; counter.test
  • Cause counter.tcl ::counter::histHtmlDisplayRow calls ncgi::value and html::font but never requires either package.
  • Prototype prototypes/1011102fff.patch (22 lines, test counter-html-1.0). Requires both packages lazily inside histHtmlDisplayRow, and counter.man mentions it. aku's 2004 idea of splitting the HTML code into its own package would be the bigger alternative.
  • Suggested action apply prototype
  • Suggested ticket comment Still present on trunk. Proposed patch loads html and ncgi on demand in histHtmlDisplayRow, so the base package keeps no hard dependency on them.
⬜1011094fff2004Restrictions for counter names not documentedpatchREPRODUCED (9.1, 8.6): counter::init ::foo and counter::init a::b fail with "parent namespace doesn't exist". counter::init a(b) fails with "…

Open ticket 1011094fff · Bug · counter · filed 2004-08-17

Full verdict

REPRODUCED (9.1, 8.6): counter::init ::foo and counter::init a::b fail with "parent namespace doesn't exist". counter::init a(b) fails with "variable isn't array".

  • Module counter 2.0.6
  • Tested tclsh9.1, tclsh8.6, direct repro script; counter.test
  • Cause counter.tcl uses the tag in variable names (counter::T-$tag, H-$tag, …), so :: and a trailing (…) break them.
  • Prototype prototypes/1011094fff.patch (27 lines, tests counter-1.2, counter-1.3). counter::init now rejects such names with a clear error, and counter.man documents the restriction.
  • Suggested action apply prototype (or the doc part only)
  • Suggested ticket comment Still true on trunk. Counter names are used inside variable names, so :: and a trailing (x) fail with obscure errors. Proposed patch: reject such names in counter::init with a clear message and document the restriction.
⬜1011096fff2004secret undocumented switches to counter::getpatchREPRODUCED (9.1, 8.6): counter::get foo -maxPerMinute on a -hist counter gives "can't read counter(maxPerMinute)". The options are still undocu…

Open ticket 1011096fff · Bug · counter · filed 2004-08-17

Full verdict

REPRODUCED (9.1, 8.6): counter::get foo -maxPerMinute on a -hist counter gives "can't read counter(maxPerMinute)". The options are still undocumented and missing from the "Invalid option" error message.

  • Module counter 2.0.6
  • Tested tclsh9.1, tclsh8.6; counter.test
  • Cause counter.tcl ::counter::get reads counter(maxPer*) without checking the type. Only -timehist counters set those values.
  • Prototype prototypes/1011096fff.patch (42 lines, tests counter-maxPer-1.0/1.1). Gives a clear error for non-timehist counters, lists the options in the error message and documents them in counter.man.
  • Suggested action apply prototype
  • Suggested ticket comment Still present. Proposed patch documents -maxPerMinute/-maxPerHour/-maxPerDay and gives a clear error for counters that are not -timehist.
⬜991452ffff2004ftp::Quote should join argumentspatchREPRODUCED (9.1, 8.6)

Open ticket 991452ffff · Bug · ftp · filed 2004-07-15

Full verdict

REPRODUCED (9.1, 8.6)

  • Module ftp 2.4.14
  • Tested ftp::Quote $h SITE {CHMOD 644} {my file} sends SITE {CHMOD 644} {my file} to the server.
  • Cause ftp.tcl ::ftp::Quote uses the list $args as the command string.
  • Prototype prototypes/991452ffff.patch (2 code lines + 2 doc lines, test ftp-quote-1.0). Uses join $args " ", as suggested in the ticket, and ftp.man says the arguments are joined with spaces.
  • Suggested action apply prototype
  • Suggested ticket comment Confirmed: arguments with spaces reach the server brace-quoted. The patch joins them with spaces as suggested, and adds a test and a doc note.
⬜862638ffff2003can't read &quot;ftp(Wait)&quot;: no such variable - background errorpatchREPRODUCED (9.1, 8.6)

Open ticket 862638ffff · Bug · ftp · filed 2003-12-19

Full verdict

REPRODUCED (9.1, 8.6)

  • Module ftp 2.4.14
  • Tested async Get (Open -command) where the 226 reply arrives before the data EOF, then Close. About 1 s later: bgerror can't read "ftp(Wait)": no such variable. Before Close, the same stray timer fires a spurious timeout callback.
  • Cause ::ftp::CopyNext re-arms the Timeout timer even for the final chunk. In async mode nothing cancels it once the reply has already been handled. ::ftp::Timeout also assumes the connection still exists.
  • Prototype prototypes/862638ffff.patch (12 changed lines in ftp.tcl, test ftp-async-timeout-1.0). CopyNext re-arms only while data remains, or in sync mode. Timeout returns if the connection is gone. Sync and async ftpd transfers still pass.
  • Suggested action apply prototype
  • Suggested ticket comment Reproduced with an async Get whose final reply arrives before the data EOF: CopyNext re-arms the timeout after the last chunk and nothing cancels it. The patch stops re-arming at the end of the transfer and guards Timeout.
🔧815405ffff2003pop3::top Hangs under wish/tkconpatchREPRODUCED (9.1, 8.6), for the "100% CPU" part. The Windows 2000/Tcl 8.4.2 wish specifics are not reproducible here: 50 TOP calls under wish on Lin…

Open ticket 815405ffff · Bug · pop3 · filed 2003-09-30

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in ead1e7510c on pop3-eof-busy-loop, not merged: Fix 815405ffff: pop3::top spun forever at full CPU when the server closed the connection in the middle of a reply, and pop3 send failed with an empty message.
  • 💬 2026-11-16 serhiy.storchaka: The CPU spin is real: RetrSlow, used by top, loops forever if the server closes the connection before the final ".", and send then fails with an empty message. Proposed fix in branch pop3-eof-busy-loop: both raise "connection closed by server". Tests pop3-ticket-815405-1 and -2.
Full verdict

REPRODUCED (9.1, 8.6), for the "100% CPU" part. The Windows 2000/Tcl 8.4.2 wish specifics are not reproducible here: 50 TOP calls under wish on Linux work.

  • Module pop3 1.11
  • Tested a fake server that closes the connection in the middle of a TOP reply makes pop3::top spin forever at ~75–100% CPU (pop3.test hangs) on 9.1 and 8.6. After a connection drop, pop3::send fails with an empty message ("POP3 STAT ERROR: "), which matches the reporter's "unknown (no error string) error out of pop3::send".
  • Cause pop3.tcl ::pop3::RetrSlow loops on gets without checking for EOF, so after the server closes, gets returns "" forever. ::pop3::send also ignores EOF.
  • Prototype prototypes/815405ffff.patch (40 changed lines, tests pop3-ticket-815405-1/-2). RetrSlow and send now raise "connection closed by server" at EOF. Without the fix, test -1 hangs (busy loop) and test -2 returns an empty error, on 9.1 and 8.6. The full pop3 suite passes on both (46/47, 1 TLS skip).
  • Suggested action apply prototype; then close (the original Windows/8.4 trigger is long obsolete)
  • Suggested ticket comment The CPU spin is real: RetrSlow (used by top) loops forever if the server closes the connection before the final ".", and send then fails with an empty message. A proposed fix with tests is ready.
⬜783878ffff2003FTP library responding WAY to slow on ftp connect timeout.patchREPRODUCED (9.1; 8.6 by code, the test needs socket -backlog)

Open ticket 783878ffff · Bug · ftp · filed 2003-08-06

Full verdict

REPRODUCED (9.1; 8.6 by code, the test needs socket -backlog)

  • Module ftp 2.4.14
  • Tested a local listener with -backlog 1 and a full queue, so SYNs are dropped and connects hang. ftp::Open ... -timeout 1 returns -1 only after 2m13s.
  • Cause ftp.tcl OpenControlConn uses a blocking socket connect, which ignores -timeout.
  • Prototype prototypes/783878ffff.patch (28 changed lines in ftp.tcl, test ftp-open-timeout-1.0, constraint backlog so it is skipped on 8.6). In synchronous mode it connects with -async and waits for writable or the timeout. Async (-command) mode is left unchanged. Refused-connection, unknown-host and ftpd transfer checks still pass.
  • Suggested action apply prototype
  • Suggested ticket comment Still present: the control connection uses a blocking connect, so -timeout is ignored and an unreachable host takes about 2 minutes. The patch connects asynchronously and honours -timeout.
⬜715193ffff2003unwanted PCDATA space trimpatchREPRODUCED (9.1, 8.6)

Open ticket 715193ffff · Bug · htmlparse · filed 2003-04-04

Full verdict

REPRODUCED (9.1, 8.6)

  • Module htmlparse 1.2.3
  • Tested the ticket's example on 9.1 and 8.6 gives RR instead of RR , and drops the single space between </b> and <i>.
  • Cause htmlparse.tcl 2treeCallback runs string trim on all text. (The set text [mapEscapes ...] next to it is dead code.)
  • Prototype prototypes/715193ffff.patch (71 lines; drops the trim and the dead code, and documents that text is kept as is; test htmlparse-${impl}-5.3 in htmlparse.tree_testsuite). The test fails without the fix. The 59 tests pass on 9.1 and 8.6. Behaviour change: whitespace-only text between tags (e.g. a newline after </title>) now becomes PCDATA nodes. This is what the reporters of this ticket and 3091244 asked for, but a human must decide whether it is acceptable.
  • Suggested action decide on the behaviour change; if accepted, apply the prototype and close 3091244 with this ticket
  • Suggested ticket comment Still present: 2treeCallback trims all PCDATA. Proposed fix: keep the text unchanged, as suggested in 3091244. This changes the tree for documents with whitespace between tags.
⬜548540ffff2002ftp::Put success on control timeoutpatchREPRODUCED (9.1, 8.6)

Open ticket 548540ffff · Bug · ftp · filed 2002-04-25

Full verdict

REPRODUCED (9.1, 8.6)

  • Module ftp 2.4.14
  • Tested fake server (dialog) that never answers a command. ftp::Delete with -timeout 1 returns 1, a false success. With the server not answering RETR, ftp::Get -variable throws can't read "ftp(Start_Time)" from ElapsedTime, which is the crash in the 2007 comment and in ticket 1709566.
  • Cause ftp.tcl ::ftp::Timeout sets ftp(state.control) 1, so WaitOrTimeout reports success.
  • Prototype prototypes/548540ffff.patch (2 changed lines in ftp.tcl, tests ftp-timeout-1.0, ftp-timeout-1.1). Close after the timeout no longer hangs. Real transfers via ftpd still pass (active and passive).
  • Suggested action apply prototype, close 1709566 as duplicate
  • Suggested ticket comment Still present: Timeout sets state.control to 1, so a timed-out command reports success, and Get then fails in ElapsedTime. The patch sets it to 0 and adds tests; Close no longer hangs afterwards.
⬜547331ffff2002stooop redefines [::proc] !REPRODUCED (9.1, 8.6; still by design)

Open ticket 547331ffff · Bug · stooop · filed 2002-04-23

Full verdict

REPRODUCED (9.1, 8.6; still by design)

  • Module stooop 4.4.2
  • Tested after package require stooop, info commands ::_proc → ::_proc and info procs ::proc → ::proc.
  • Cause stooop.tcl line 18 rename proc _proc, then defines ::proc. stooop needs this so that proc Class::member {this ...} works transparently.
  • Prototype none (needs the opt-in API discussed in 2002, i.e. stooop::import / stooop::proc, and a compatibility decision)
  • Suggested action human decision. Either implement opt-in replacement of ::proc (breaks existing users), or close as Won't Fix and document the global side effect.
  • Suggested ticket comment Still the case in stooop 4.4.2: loading it replaces ::proc and creates ::_proc. Fixing it needs an opt-in API and breaks current users. Should we document it and close?
⬜509776ffff2002Locale dependant test failurespatchREPRODUCED (8.6 for mime and many other modules; 9.1 and 8.6 for doctools2base). The argv0 part was fixed in 2003.

Open ticket 509776ffff · Bug · None · filed 2002-01-28

Full verdict

REPRODUCED (8.6 for mime and many other modules; 9.1 and 8.6 for doctools2base). The argv0 part was fixed in 2003.

  • Module test harness (support/devel/all.tcl), mime, doctools2base
  • Tested the full suite with tclsh8.6 under LANG=C vs uk_UA.koi8u. Then module runs under C, C.UTF-8, uk_UA.koi8u and pl_PL.iso88592 on 8.6 and 9.1.
    • 8.6 in a non-UTF-8 locale: mime fails 7 (C) or 9 (koi8u, iso8859-2) tests. textutil 18, base64 2, yaml 2, fumagic 3, report 2 and struct 2 fail too. All pass under C.UTF-8.
    • 9.1: mime passes everywhere.
    • doctools-msgcat-2.0 fails under any non-English locale on both 8.6 and 9.1.
  • Cause:
    • (1) support/devel/all.tcl sources test files with plain source, so on 8.6 the UTF-8 test files are read in the system encoding.
    • (2) doctools2base/msgcat.test resets env(LANG) after msgcat has already picked its locale, so doctools::msgcat::init idx looks for the uk/pl catalogs.
  • Prototype prototypes/509776ffff.patch (32 lines): source -encoding utf-8 in all.tcl, and msgcat::mclocale c in msgcat.test. The existing tests serve as the regression tests: mime 4.4/5.x/6.2/7.1, textutil, base64, yaml and doctools-msgcat-2.0 fail without the fix and pass with it under C/koi8u/iso8859-2 on 8.6 and 9.1.
    • Still failing on 8.6 in the C locale even with the fix: fumagic 3, report-28.13, matrix-5.8. These are probably channel output encoding issues, not investigated.
    • Alternative for mime only: replace the non-ASCII literals in mime.test with \u escapes (tried, works; it also helps standalone tclsh mime.test).
  • Suggested action apply the prototype. Look at the remaining fumagic/report/matrix failures separately.
  • Suggested ticket comment The argv0 part was fixed long ago. The locale part is still real on 8.6: test files are sourced in the system encoding, and the msgcat test ignores the already-chosen locale. A harness fix (source -encoding utf-8) and a msgcat.test fix are proposed.
⬜472259ffff2001getoptions -- help or error?patchREPRODUCED (9.1, 8.6)

Open ticket 472259ffff · Bug · cmdline · filed 2001-10-17

Full verdict

REPRODUCED (9.1, 8.6)

  • Module cmdline 1.5.3
  • Tested -?, -help, -bogus and -f (missing argument) all give the same message and the same CMDLINE USAGE errorcode. The caller cannot tell help from a real error, and the bad option is not named. The same is true for getKnownOptions and typedGetoptions.
  • Cause cmdline.tcl:getoptions/getKnownOptions set result(?) on any getopt error. typedGetoptions throws the plain usage error. The specific message ("Illegal option ...") is discarded.
  • Prototype prototypes/472259ffff.patch (144 lines, tests cmdline-8.1, cmdline-8.2, typed-cmdline-8.1, plus a cmdline.man paragraph). This is the compatible variant: the message stays the same, the help case keeps CMDLINE USAGE, and invalid input gets CMDLINE USAGE INVALID <specific message>. trap {CMDLINE USAGE} still catches both. All tests pass on 9.1 and 8.6 (139 passed, 9 skipped). The 3 new tests fail without the fix. Prefixing the message itself with the problem would break 8 existing tests, so it was not done.
  • Suggested action decide on the errorcode extension, then apply prototype
  • Suggested ticket comment Still true. Help and invalid options throw the same CMDLINE USAGE error. A compatible fix keeps CMDLINE USAGE for -?/-help and adds INVALID <message> to the errorcode for real errors.
🔧450914ffff2001example access Tk internalspatchREPRODUCED (Tk 9.0, 9.1)

Open ticket 450914ffff · Bug · ftp · filed 2001-08-14

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in a3dab9327c on ftp-hpupdate-tk9, not merged: Fix 450914ffff: the ftp hpupdate example failed at startup on Tk 9: it exposed the private tkButtonInvoke command, which is gone. Use a small helper that flas
  • 💬 2026-11-15 serhiy.storchaka: tkButtonInvoke is gone in Tk 9, so the hpupdate example now fails at startup. Proposed fix in branch ftp-hpupdate-tk9: a small helper flashes and invokes the button instead.
Full verdict

REPRODUCED (Tk 9.0, 9.1)

  • Module examples/ftp/hpupdate.tcl
  • Tested wish9.1/9.0/8.6 on Xvfb :107. On Tk 9.x, startup fails with invalid command name "::tk::unsupported::ExposePrivateCommand" (hpupdate.tcl line 36). It still works on 8.6.
  • Cause the 2001 workaround exposes the private tkButtonInvoke via ::tk::unsupported::ExposePrivateCommand, and Tk 9 removed that command.
  • Prototype prototypes/450914ffff.patch (22 changed lines). It adds a local ButtonInvoke proc (flash + invoke, skipped when disabled) used by the 6 key bindings. No regression test: examples/ has no test suite. Checked by hand on Xvfb with 9.1 and 8.6: the app starts, the bindings use the new proc, and invoking it fires the button command.
  • Suggested action apply prototype, close as Fixed
  • Suggested ticket comment The workaround stopped working in Tk 9 (::tk::unsupported::ExposePrivateCommand is gone). The patch replaces tkButtonInvoke with a small public-API helper (flash + invoke).

Partly fixed 22

🔧cdbdd399e42025package require rest does not work with tcl8.6patchPARTLY-FIXED. The reported package require Tcl 8.5 9+ (invalid version syntax, added in 0b3dd048f8 2025-07-16) was fixed on trunk in [f14416cf2…

Open ticket cdbdd399e4 · Bug · rest · filed 2025-12-03

🔧 Since triage (updated 2026-11-16)

Status now: Open

  • 🔧 2026-11-15 check-in d828c0f28f on pkgindex-version-fixes, not merged: Fix 61e49798ee, 523f7e0556, cdbdd399e4: pkgIndex.tcl entries that do not match the code. nettool's index sourced nexttool.tcl, so package require nettool
  • 💬 2026-11-15 serhiy.storchaka: The Tcl 9-only requirement was fixed on trunk in f14416cf2a. However, package require rest still fails on trunk with every Tcl version, because pkgIndex.tcl says 1.7 while rest.tcl provides 1.8. Proposed fix in branch pkgindex-version-fixes, with a test.
Full verdict

PARTLY-FIXED. The reported package require Tcl 8.5 9+ (invalid version syntax, added in 0b3dd048f8 2025-07-16) was fixed on trunk in f14416cf2a (2025-08-19). Tcllib 2.0 itself shipped rest 1.7 with "8.5 9" and was never affected; the reporter had a trunk snapshot. On trunk package require rest still fails on all Tcl versions: "attempt to provide package rest 1.7 failed: package rest 1.8 provided instead". pkgIndex.tcl and rest.man were not bumped along with rest.tcl. sak validate versions rest flags it.

  • Module rest 1.8
  • Tested auto_path to the source tree, with a stub tdom (no tdom installed). Fails on 8.6 and 9.1; works after the fix.
  • Cause modules/rest/pkgIndex.tcl says 1.7 while rest.tcl provides 1.8 (since 0b3dd048f8).
  • Prototype prototypes/cdbdd399e4.patch (56 lines; pkgIndex and rest.man to 1.8; new rest.test, test rest-0.1, which checks that the index and provide versions agree). This completes an existing bump rather than adding one. The test fails without the fix and passes on 9.1 and 8.6.
  • Suggested action apply prototype, then close as Fixed
  • Suggested ticket comment The "9+" requirement was fixed in f14416cf2a. Trunk still has pkgIndex.tcl at 1.7 vs rest.tcl 1.8, which makes package require rest fail; proposed fix syncs them.
⬜8f0ff730832024tcllibc and sha256c leads to error "Headers for API sha256c not found in"PARTLY-FIXED (692adc1efb, 2024-09-17, in tcllib 2.0)

Open ticket 8f0ff73083 · Bug · sha1 · filed 2024-09-17

Full verdict

PARTLY-FIXED (692adc1efb, 2024-09-17, in tcllib 2.0)

  • Module sha1 / sha256 1.0.6, sha256c
  • Tested code reading only. critcl is not installed here (CANNOT-TEST).
  • Cause sha2::LoadAccelerator now tries sha256c before tcllibc (692adc1efb). If other code loads tcllibc first, the tcllibc shared library has already run package provide sha256c, so the sha256c.tcl critcl definitions (critcl::cheaders) never run. The reporter's last comment says the package forget sha256c workaround then fails with "does not support stubs". aku explains why the provide call in tcllibc cannot simply be dropped.
  • Prototype none (critcl/stub-table packaging design question, cannot test here)
  • Suggested action keep open for aku (design), or close as Fixed for the search-order part if the remaining load-order case is accepted as a limitation.
  • Suggested ticket comment The search-order change 692adc1efb is in 2.0. The remaining case, where tcllibc was loaded before sha256c by someone else, needs a packaging decision (how tcllibc exposes the sha256c stub headers).
⬜f70f7ea5142023tcllib 1.21 / Tcl 8.6.13: Test errors on windowsPARTLY-FIXED: on trunk with Tcl 9.1 on Windows, the fileutil, fumagic, math, mime, report, tar, textutil, virtchannel_base and yaml failures of the…

Open ticket f70f7ea514 · Bug · (unused) · filed 2023-08-27

Full verdict

PARTLY-FIXED: on trunk with Tcl 9.1 on Windows, the fileutil, fumagic, math, mime, report, tar, textutil, virtchannel_base and yaml failures of the report are gone. Still failing on Windows only: httpd (4), irc picoirc (20), pop3d (1), profiler (1) and uri (1). crc (7) and hook (2) fail on 9.1 on Linux too.

  • Module many (fileutil, fumagic, httpd, irc, math, mime, profiler, report, struct, tar, textutil, uri, virtchannel_base, yaml)
  • Tested none (Linux only); timeline review
  • Cause various platform assumptions in the tests (path separators, CRLF, timezones)
  • Prototype none
  • Suggested action rerun the full suite on Windows with trunk. Close this umbrella ticket once pbrooks's branches are merged, filing per-module tickets for anything left.
  • Suggested ticket comment Many of these Windows test failures were fixed on trunk in 2026 (fileutil, tar, virtchannel, profiler pending). A fresh Windows run of trunk is needed to see what remains.
Windows VM, run 1
  • Verdict PARTLY-FIXED: on trunk with Tcl 9.1 on Windows, the fileutil, fumagic, math, mime, report, tar, textutil, virtchannel_base and yaml failures of the report are gone. Still failing on Windows only: httpd (4), irc picoirc (20), pop3d (1), profiler (1) and uri (1). crc (7) and hook (2) fail on 9.1 on Linux too.
  • Tested full sak test run of trunk with tclsh91 on the Windows VM (websocket.test left out because it contacts an external server).
  • Cause:
    • picoirc-2.x/3.x/4.x: the sent lines end in \x0D; the test channel keeps the CR of the CRLF translation on Windows.
    • uri-4.1: uri::geturl file://C:/... drops the drive letter (:/Users/...).
    • httpd-cgi-0001/0002 (504) and httpd-proxy-0004/0005 (empty reply): CGI and proxy subprocesses are not run on Windows.
    • pop3-srv-5.0: the greeting does not match the expected pattern.
    • profiler-11.0-tkt-3603260fff: the timing test; pbrooks's branch pbrooks-2026-11-windows-profiler-fixes addresses it.
  • Tcl 8.6.18 on the same VM, whose ANSI code page is Cyrillic (cp1251): the same Windows-only failures, plus base64/yencode (2), fumagic (3), mime (9), report (2), struct matrix (2), textutil (18) and yaml json2huddle (2). Almost all of these come from test files being sourced in the system encoding on 8.6, e.g. tr\u0413\xA9s for très. The prototype for 509776ffff (source test files as UTF-8) is the fix. fumagic.filetype-1.25.1/1.27 recognise the PE test file only as MS-DOS on Windows.
  • Suggested action close this umbrella ticket and file per-module tickets for picoirc (CR in the test harness), uri (file:// with drive letters) and httpd CGI/proxy on Windows. Merge the profiler branch.
⬜8bb4e24bc12023mime.tcl some multipart/mixed messages not parsed correctly. Errorhandling, WhitespacePARTLY-FIXED (the actual bug, a stray "]", was fixed in 42266acec4 on 2023-08-16, in tcllib 2.0. The rest is a refactor of error handling and whi…

Open ticket 8bb4e24bc1 · Bug · mime · filed 2023-08-16

Full verdict

PARTLY-FIXED (the actual bug, a stray "]", was fixed in 42266acec4 on 2023-08-16, in tcllib 2.0. The rest is a refactor of error handling and whitespace for Jim Tcl, not a bug.)

  • Module mime 1.7.2
  • Tested code check (the bracket is gone on trunk)
  • Cause fixed
  • Prototype none
  • Suggested action close as Fixed; the Jim-compatibility refactor could be a separate RFE if wanted
  • Suggested ticket comment The parsing bug was fixed in 42266acec4 (tcllib 2.0). The error-handling and whitespace rework is a separate cleanup; closing this as fixed.
⬜1f29f7baf92023ncgi::encode broken with Tcl 9 / ncgi::encode encodes to iso8859-1patchPARTLY-FIXED. The Tcl 9 / TIP 465 encode "(" error is fixed (7804eb94e9 2023-10-06, merged 758418619e 2023-11-14, tcllib 2.0; test ncgi-4.5). Sti…

Open ticket 1f29f7baf9 · Bug · ncgi · filed 2023-07-07

Full verdict

PARTLY-FIXED. The Tcl 9 / TIP 465 encode "(" error is fixed (7804eb94e9 2023-10-06, merged 758418619e 2023-11-14, tcllib 2.0; test ncgi-4.5). Still broken on 9.1 and 8.6: any character above U+00FF raises "can't read map(й): no such element in array", U+0080–U+00FF are encoded as Latin-1 (ö gives %F6), U+0100 gives the bogus %100, and NUL is not encoded.

  • Module ncgi 1.4.6
  • Tested ncgi::encode with й, 中, U+1F600, ö, NUL, on 9.1 and 8.6
  • Cause ncgi.tcl:encode maps characters through map, which is built for code points 1..256
  • Prototype prototypes/1f29f7baf9.patch (105 lines, about 36 changed. Converts to UTF-8 before mapping, fixes the table to 0..255, and documents this in ncgi.man. New tests ncgi-4.6..4.9 fail without the fix. decode still accepts the old Latin-1 escapes, so old encoded data decodes the same.)
  • Suggested action apply prototype. Human decision: ö changes from %F6 to %C3%B6. That matches http::formatQuery, browsers and RFC 3986, but it is a visible behaviour change.
  • Suggested ticket comment The Tcl 9 error is fixed in tcllib 2.0, but encode still failed for any character above U+00FF. It now encodes UTF-8, like http::formatQuery.
⏳d01de3281f2022websocket client request has two Connection headersPARTLY-FIXED (fixed with http 2.10 in Tcl 9.0/9.1; still sent twice with http 2.9.9 in Tcl 8.6)

Open ticket d01de3281f · Bug · websocket · filed 2022-06-30 · Pending (resolved, waiting for confirmation)

⏳ Since triage (updated 2026-11-16)

Status now: Pending (Fixed)

    Full verdict

    PARTLY-FIXED (fixed with http 2.10 in Tcl 9.0/9.1; still sent twice with http 2.9.9 in Tcl 8.6)

    • Module websocket 1.6 (uses Tcl's http)
    • Tested local 127.0.0.1 socket server that logs the request headers, then websocket::open. 9.1.1 and 9.0.5 (http 2.10.3) send one header, Connection: upgrade, close. 8.6.18 (http 2.9.9) sends Connection: close and Connection: Upgrade.
    • Cause http 2.9.9 geturl forces -keepalive 0 for upgrade requests and writes its own Connection: close line, followed by the user-supplied Connection: Upgrade header. websocket cannot avoid this from tcllib.
    • Prototype none (the remaining problem is in Tcl 8.6's http package)
    • Suggested action close as Fixed (Tcl 9 http). Optionally ask whether http 2.9.x in 8.6 should get the header merge, as kjnash suggested.
    • Suggested ticket comment With http 2.10 (Tcl 9.0+) the request carries a single "Connection: upgrade, close" header. Tcl 8.6's http 2.9.9 still sends two Connection headers; that part belongs to Tcl's http package, not tcllib.
    ⬜827244d3ee2016get_ip_addresses is deprecated and missing in latest twapiPARTLY-FIXED: without twapi, nettool::ip_list works on Windows 9.1 (returns the VM's two IPv4 addresses). package require nettool itself fails …

    Open ticket 827244d3ee · Bug · nettool · filed 2016-05-17

    Full verdict

    PARTLY-FIXED: without twapi, nettool::ip_list works on Windows 9.1 (returns the VM's two IPv4 addresses). package require nettool itself fails (pkgIndex typo, 61e49798ee), so nettool.tcl was sourced directly. The twapi variants, which call the undefined ldelete, could not be tested (twapi not installed).

    • Module nettool 0.5.4 (platform_windows_twapi section)
    • Tested code reading. The ip_list variant for twapi checks for ::twapi::get_ip_addresses, and if it is missing, falls back to ::twapi::get_system_ipaddrs (-version 4 or -ipversion 4). This is in the tcllib-1-19 tag and not in tcllib-1-18. NEEDS-WINDOWS to run it.
    • Cause every ip_list variant calls ldelete result 127.0.0.1, and nettool never defines ldelete. On Linux ip_list fails with invalid command name "ldelete", which I reproduced. The twapi variants should fail the same way.
    • Prototype none separately. The ldelete definition is part of prototypes/9a21637273.patch (test ip-list-ldelete).
    • Suggested action close as Fixed for the API rename, once the ldelete fix from 9a21637273 is applied (or keep open until then).
    • Suggested ticket comment Since tcllib 1.19, nettool falls back to twapi::get_system_ipaddrs when get_ip_addresses is missing. ip_list still fails on every platform because ldelete is undefined. That is fixed together with 9a21637273.
    Windows VM, run 1
    • Verdict PARTLY-FIXED: without twapi, nettool::ip_list works on Windows 9.1 (returns the VM's two IPv4 addresses). package require nettool itself fails (pkgIndex typo, 61e49798ee), so nettool.tcl was sourced directly. The twapi variants, which call the undefined ldelete, could not be tested (twapi not installed).
    ⬜042d5837382016Documentation bugs - oometa, tool, cluster, ooutil, processman, trypatchPARTLY-FIXED. Fixed: try/throw (4e8a0be16c) and tcldes/tclDESjr (7f225e0906) in 2016. oometa and udpcluster (formerly nameserv::cluster) now have p…

    Open ticket 042d583738 · Bug · Documentation · filed 2016-02-23

    Full verdict

    PARTLY-FIXED. Fixed: try/throw (4e8a0be16c) and tcldes/tclDESjr (7f225e0906) in 2016. oometa and udpcluster (formerly nameserv::cluster) now have pages. tool.man now says "TclOO Library (TOOL) Framework", and "Dictionary Tools" is the legit tool::dict_ensemble page. Still open: processman titledesc is still the cron text, and modules/tool/meta.man is still a stale copy of ooutil.man (oo::util 1.2.2 vs 1.2.3), so the TOC lists oo::util twice.

    • Module oometa, tool, udpcluster, ooutil, processman, try (docs)
    • Tested grepped the .man sources and embedded/md/toc.md on trunk
    • Prototype prototypes/042d583738.patch (1 line, docs only, no test possible): processman titledesc becomes "Platform-independent management of child processes". Removing tool/meta.man (or making it document oo::meta) is left for a human.
    • Suggested action apply the processman fix; delete modules/tool/meta.man (duplicate) and regenerate the docs; then close
    • Suggested ticket comment Most items are fixed. Still open: the processman titledesc is copied from cron, and tool/meta.man is a stale duplicate of ooutil.man, so oo::util appears twice in the TOC.
    ⬜561c505d972015WebSocket headers not correctly checkedpatchPARTLY-FIXED (Connection check added in aa43bc825c, tcllib 1.21; the Upgrade check is still missing). REPRODUCED (9.1, 8.6): websocket::validate …

    Open ticket 561c505d97 · Bug · websocket · filed 2015-04-16

    Full verdict

    PARTLY-FIXED (Connection check added in aa43bc825c, tcllib 1.21; the Upgrade check is still missing). REPRODUCED (9.1, 8.6): websocket::validate accepts a request that has no Upgrade header at all. An Upgrade header without websocket is rejected.

    • Module websocket 1.6
    • Tested websocket::validate {Connection Upgrade Sec-WebSocket-Version 13 Sec-WebSocket-Key ...} returns a dict instead of raising an error.
    • Cause websocket.tcl ::websocket::validate: $websocket is checked only inside the upgrade branch of the header loop, never after the loop.
    • Prototype prototypes/561c505d97.patch (54 lines, tests websocket-validate-1.0, -1.1). It adds the check after the loop, as in the ticket's patch. The new tests pass on 9.1 and 8.6, and websocket-validate-1.0 fails without the fix. I ran them in a copy of the test file without the network tests, because websocket.test talks to ws.ifelse.io.
    • Suggested action apply prototype
    • Suggested ticket comment The Connection check was added for 1.21, but a request without any Upgrade header is still accepted. Proposed patch adds the missing check, with a test.
    ⬜3613162fff2013Adding support to log save to a filePARTLY-FIXED (writing to a file already works through logger::appender::fileAppend -outputChannel, d01676c313, 2005-11-14, but it is undocumented…

    Open ticket 3613162fff · RFE · log · filed 2013-05-13

    Full verdict

    PARTLY-FIXED (writing to a file already works through logger::appender::fileAppend -outputChannel, d01676c313, 2005-11-14, but it is undocumented. Rotation is not implemented)

    • Module logger 0.9.5, logger::appender 1.4
    • Tested logger::utils::applyAppender -appender fileAppend -serviceCmd $log -appenderArgs [list -outputChannel $f] writes [date] [foo] [global] [warn] to file to the file on 9.1 and 8.6. fileAppend is missing from loggerAppender.man.
    • Cause n/a. The attached logger.patch is a reversed p4 diff (it removes the feature). It has no rotation and calls fileutil::appendToFile for every line.
    • Prototype none (the patch is unusable. Rotation is a new feature)
    • Suggested action close as Out of date and point to fileAppend. Separately, document fileAppend in loggerAppender.man (DOCS).
    • Suggested ticket comment Logging to a file is possible with the fileAppend appender (-outputChannel $chan), which needs documenting. The attached patch is reversed and has no rotation; rotation can be done by reopening the channel.
    ⬜3138526fff2010Wrong email server name gives unrelated error messagepatchPARTLY-FIXED (9.1, 8.6)

    Open ticket 3138526fff · Bug · smtp · filed 2010-12-16

    Full verdict

    PARTLY-FIXED (9.1, 8.6)

    • Module mime / smtp 1.5.2
    • Tested a name that cannot be resolved (simulated socket error) gives a clear couldn't open socket: .... A closed local port gives error reading "sockNN": transport endpoint is not connected (9.1) or socket is not connected (8.6). A host that accepts and closes gives "premature end-of-file from server" (accurate).
    • Cause smtp.tcl: readable. After a failed socket -async connect, gets reports ENOTCONN. The real error ("connection refused") is only in fconfigure -error, which is never queried. On 8.5, gets even returns -1 without eof, so the connect wait runs until the timeout.
    • Prototype prototypes/3138526fff.patch (136 lines, mostly new smtp.test; tests smtp-2.1, smtp-2.2). Checks fconfigure -error first in readable. mime tests: 104/104 on 9.1 and 8.6. smtp-2.1 fails without the fix.
    • Suggested action apply prototype
    • Suggested ticket comment Unknown host names now give "couldn't open socket: ...". A refused connection still gave a misleading "not connected" error. Proposed fix with tests in prototypes/3138526fff.patch.
    ⬜3058654fff2010Incorrect nroff generationpatchPARTLY-FIXED (.CS/.CE for examples and whitespace compaction were added in 2010, ChangeLog [Bug 3058654]). REPRODUCED (9.1, 8.6): a multi-line [tit…

    Open ticket 3058654fff · Bug · doctools · filed 2010-09-03

    Full verdict

    PARTLY-FIXED (.CS/.CE for examples and whitespace compaction were added in 2010, ChangeLog [Bug 3058654]). REPRODUCED (9.1, 8.6): a multi-line [titledesc]/[moddesc] still gives a multi-line NAME line and a broken .TH line. The NAME-lists-commands part is still an open RFE.

    • Module doctools 1.6.1 (mpformats/fmt.nroff)
    • Tested formatted a page with [titledesc {A long\n multi-line description}] to nroff on 9.1/8.6. NAME is spread over 2 lines, and the moddesc in .TH is cut at the newline.
    • Cause fmt.nroff fmt_manpage_begin compacts only [ \t]+, not newlines. Also, the loop variable text stays upvar-linked to the last variable, so later set text [c_held precomments] overwrites it. That is harmless today (copyright is used before), but a trap when extending the loop.
    • Prototype prototypes/3058654fff.patch (+13 lines, test doctools-8.5). It compacts all whitespace in module/shortdesc/description, using a separate link variable. The test fails on 9.1 and 8.6 without the fix. doctools passes 600/600 on both.
    • Suggested action apply the prototype. Keep the "NAME should list the commands" part open as an RFE, or split it into a new ticket.
    • Suggested ticket comment Examples use .CS/.CE since 2010, but newlines in titledesc/moddesc still go into the NAME and .TH lines. Proposed fix with test. Listing the commands in NAME is still not done.
    ⬜2928355fff2010ftp::Rename causes ftpd to crashpatchPARTLY-FIXED (the missing upvar was fixed in 2010, ChangeLog 2010-01-20, tcllib 1.13). Renaming is still broken on 9.1 and 8.6.

    Open ticket 2928355fff · Bug · ftpd · filed 2010-01-08

    Full verdict

    PARTLY-FIXED (the missing upvar was fixed in 2010, ChangeLog 2010-01-20, tcllib 1.13). Renaming is still broken on 9.1 and 8.6.

    • Module ftpd 1.4.1
    • Tested ::ftpd::fsFile::fs rename with a docRoot that is not the process cwd. The target name is not joined with docRoot, so the file is renamed relative to the server's cwd, or the call fails. When the target directory is missing it replies "550" and then still tries the rename, which errors, and RNTO sends a second reply "500". RNTO to a new name was also denied by the default fileAuth (keithv's comment), which is 1508656fff.
    • Cause ftpd.tcl ::ftpd::fsFile::fs rename: $newname lacks file join $docRoot, and there is no return after the 550.
    • Prototype prototypes/2928355fff.patch (on top of 34a589f660.patch; 2 lines; tests ftpd-3.0, ftpd-3.1). If 3312860fff is applied too, the rename target needs the same path handling.
    • Suggested action apply prototype together with 1508656fff, then close as Fixed. The design question in aku's comment (separate rename-src/rename-dst auth modes; RNFR still needs the source file itself to be writable) can be left out of this ticket.
    • Suggested ticket comment The crash was fixed in 2010. Renaming with the default file system still did not work: the new name was not resolved against the docRoot, and a missing target directory produced two replies. Proposed fix with tests in the prototype; the permission part is covered by [1508656].
    ⬜2054498fff2008Race condition in ftp::Get -variablePARTLY-FIXED (sync mode fixed since tcllib 1.17 / ftp 2.4.13; async -command mode still affected on 9.1, 8.6)

    Open ticket 2054498fff · Bug · ftp · filed 2008-08-16

    Full verdict

    PARTLY-FIXED (sync mode fixed since tcllib 1.17 / ftp 2.4.13; async -command mode still affected on 9.1, 8.6)

    • Module ftp 2.4.14
    • Tested a mock server that sends "226" before it writes the file data (300 ms later), active and passive. Synchronous Get -variable returns the full 8000 bytes (WaitComplete waits for the data EOF, and WaitDataConn 6474b535bb, tkt eb0b15d598, 2014-02-21 covers late data connections). In async mode (Open -command), the get callback is called with 0 bytes and the variable stays empty, because the data arrives after completion. With normal ordering (data, then 226) async works.
    • Cause ftp.tcl StateHandler get_close. In async mode the state machine completes on 226 without waiting for the data connection EOF (WaitComplete only waits when ftp(Command) is empty).
    • Prototype none. A fix needs deferred completion of the async state machine until the data EOF, for get_close, list_close and reget_close. That is more than a local fix.
    • Suggested action keep open, retitle to "async mode"
    • Suggested ticket comment The synchronous case is fixed since ftp 2.4.13 (tcllib 1.17). It still happens with ftp::Open -command: if the server sends 226 before the data, the get callback receives empty data.
    ⬜1948953fff2008ncgi: incomplete url encoding/decodingPARTLY-FIXED. decode handles 2- and 3-byte UTF-8 since ncgi 1.4/1.4.1 (a6430d4122 2012-05, b6632970f6 2013-01; tcllib 1.15). The remaining gaps are…

    Open ticket 1948953fff · Patch · ncgi · filed 2008-04-22

    Full verdict

    PARTLY-FIXED. decode handles 2- and 3-byte UTF-8 since ncgi 1.4/1.4.1 (a6430d4122 2012-05, b6632970f6 2013-01; tcllib 1.15). The remaining gaps are 4-byte sequences in decode, and encode, which fails on any character above U+00FF ("can't read map(й)") and encodes U+0080–U+00FF as Latin-1.

    • Module ncgi 1.4.6
    • Tested ncgi::encode й gives an error on 9.1 and 8.6. ncgi::decode %F0%A0%AE%B7 returns 4 Latin-1 characters.
    • Cause ncgi.tcl:encode looks up the map array, which only covers characters 1–256. decode has no 4-byte regsub.
    • Prototype none of its own. Covered by prototypes/66063baec4.patch (decode) and prototypes/1f29f7baf9.patch (encode). The attached lsflow code is not applicable as is: it lives in another namespace and emits raw hex digits for invalid bytes.
    • Suggested action close together with 66063baec4 and 1f29f7baf9 when they are applied
    • Suggested ticket comment decode got UTF-8 support in ncgi 1.4. The remaining 4-byte decode and non-Latin-1 encode issues are handled in 66063baec4 and 1f29f7baf9.
    ⬜1800338fff2007dtplite can't find it's matching tcllib if --prefix usedPARTLY-FIXED (9.1, 8.6). The "picks up an older tcllib" half is gone: apps/dtplite now does package require dtplite 1.0.5, so an older installed …

    Open ticket 1800338fff · Bug · doctools · filed 2007-09-22

    Full verdict

    PARTLY-FIXED (9.1, 8.6). The "picks up an older tcllib" half is gone: apps/dtplite now does package require dtplite 1.0.5, so an older installed tcllib cannot satisfy it. The other half is reproduced: after installer.tcl -pkg-path <dir not on auto_path> the installed dtplite fails with "can't find package dtplite 1.0.5". The same applies to all the apps (mkdoc, pt, page, nns...).

    • Module apps/dtplite (dtplite package 1.0.5), installer.tcl
    • Tested installed trunk to a private prefix with tclsh9.1, then ran inst/bin/dtplite with tclsh9.1 and tclsh8.6 without TCLLIBPATH. Both failed as described.
    • Cause installer.tcl ainstall copies apps verbatim and does not record the package install path. AK's attached auto_path snippet was never applied.
    • Prototype none. The installer could add lappend auto_path <pkg-path> to the apps it installs, but AK pointed out it conflicts with relocatable/batteries-included distributions, so the maintainer must decide. Setting TCLLIBPATH, the standard Tcl way to find extensions in a non-default location, works.
    • Suggested action maintainer decision. Either close as Won't Fix (use TCLLIBPATH or install into Tcl's lib dir), or have the installer patch the apps' auto_path when -pkg-path is not on the tclsh's auto_path.
    • Suggested ticket comment The installed apps now require their exact package version, so a stale older tcllib is no longer picked up. Installing into a directory outside Tcl's package path still needs TCLLIBPATH, because the installer does not patch the apps.
    ⬜1558351fff2006ASN LibraryPARTLY-FIXED (asn 0.8, 2007): asnPeekTag/asnTag/asnRetag handle multi-byte and non-universal tags; asnChoice/asnChoiceConstr exist for encoding. St…

    Open ticket 1558351fff · Bug · asn · filed 2006-09-14

    Full verdict

    PARTLY-FIXED (asn 0.8, 2007): asnPeekTag/asnTag/asnRetag handle multi-byte and non-universal tags; asnChoice/asnChoiceConstr exist for encoding. Status of the other points:

    • Module asn 0.8.5
      • point 3 (PeekBytes): string range is enough.
      • point 4 (getByte returns decimal): not a bug.
      • point 1 (strict universal-tag checks in asnGet*): still present, but asnRetag is the intended workaround.
      • point 5 (decoding CHOICE/SEQUENCE OF): still a vague RFE.
    • Tested code read (asn.tcl:884-990, 1036)
    • Cause n/a
    • Prototype none (catch-all RFE)
    • Suggested action close as Fixed/Out of date; open specific RFEs if anyone still needs CHOICE decoding
    • Suggested ticket comment Multi-byte and private tags are handled since asn 0.8 (asnPeekTag, asnTag, asnRetag). The remaining items are design choices or very general requests. Closing; please open a specific ticket if something concrete is still missing.
    ⬜1263071fff2005-configuremethodPARTLY-FIXED (snit 2.x accepts a command prefix; snit 1.x does not)

    Open ticket 1263071fff · Bug · snit · filed 2005-08-18

    Full verdict

    PARTLY-FIXED (snit 2.x accepts a command prefix; snit 1.x does not)

    • Module snit 2.3.4 / 1.4.3
    • Tested option -text -configuremethod [list SetText 1]. snit 2.3.4 works on 9.1/8.6 ({*} expansion). snit 1.4.3 fails with "::t1 SetText 1" is not defined.
    • Cause snit 2 RT code expands the method with {*}. snit.man still documents methodName only.
    • Prototype none (doc/design decision: the author called the old behaviour "by design")
    • Suggested action close as Fixed for snit 2 and document that a command prefix is allowed, or close as Won't fix for snit 1.
    • Suggested ticket comment Snit 2.x accepts a method name plus extra arguments for -configuremethod (and -cgetmethod), though it is not documented. Snit 1 does not.
    ⬜1193023fff2005smtp can have no originator/frompatchPARTLY-FIXED (9.1, 8.6)

    Open ticket 1193023fff · Bug · smtp · filed 2005-04-30

    Full verdict

    PARTLY-FIXED (9.1, 8.6)

    • Module mime / smtp 1.5.2
    • Tested sendmessage with tcl_platform(user) set to "" and no From, and with -header {From ""}, on 9.1 and 8.6.
    • Cause smtp.tcl: sendmessage. A default array set aprops {error "invalid address \"$from\""} was added at some point, so it no longer fails with "can't read aprops(error)". But the default is in braces, so the message shows a literal $from: error in -originator: invalid address "$from".
    • Prototype prototypes/1193023fff.patch (144 lines, mostly the new file modules/mime/smtp.test; tests smtp-1.1, smtp-1.2). Builds the default with [list ...] and the right variable. mime module tests: 104/104 pass on 9.1 and 8.6. The new tests fail without the fix. All four smtp prototypes create smtp.test with the same preamble. Applied together, merge them into one file.
    • Suggested action apply prototype
    • Suggested ticket comment The missing-originator case now gives an error, but the message shows a literal "$from". Proposed fix with tests in prototypes/1193023fff.patch.
    ⬜1005377fff2004struct::stack name collisionspatchPARTLY-FIXED (fixed for the default TclOO implementation; still REPRODUCED with stack_tcl.tcl, used when TclOO is absent, e.g. tclsh8.5, and in the…

    Open ticket 1005377fff · Bug · struct :: stack · filed 2004-08-08

    Full verdict

    PARTLY-FIXED (fixed for the default TclOO implementation; still REPRODUCED with stack_tcl.tcl, used when TclOO is absent, e.g. tclsh8.5, and in the critcl implementation)

    • Module struct::stack 1.5.4
    • Tested 9.1 and 8.6 (TclOO impl): unnamed stacks are ::oo::ObjNN, no collision. 8.5 (pure Tcl impl): with proc stack2 {} defined, struct::stack fails with command "::stack2" already exists. The "always global namespace" part was fixed long ago: names are now created in the caller's namespace.
    • Cause stack_tcl.tcl stack_tcl generates stack$counter once and errors on a clash. stack_c.tcl SDnewName does the same. queue_tcl/queue_c, tree_c and graph_tcl use the same pattern.
    • Prototype prototypes/1005377fff.patch (82 lines, test stack-tcl-1.3 with new constraint stack_tcl). Fix: skip generated names that already exist. The new test runs only for the pure Tcl impl (tclsh8.5). It fails there without the fix and passes with it. On 9.1/8.6 the struct suite is unchanged (test skipped). On 8.5 the pre-existing failure set-tcl-1.0 is unrelated. The critcl impl was not changed or tested (no tcllibc build).
    • Suggested action apply prototype; optionally do the same for queue_tcl and the C implementations
    • Suggested ticket comment With TclOO (Tcl 8.6+) unnamed stacks are TclOO objects and don't collide. The pure-Tcl fallback and the critcl implementation still fail if stackN already exists. Proposed fix for the Tcl fallback: skip existing names (patch with test).
    ⬜744276ffff2003ftp::Get of a nonexistent filepatchPARTLY-FIXED (9.1, 8.6). The reported passive-mode unset ftp(state.data) error was fixed in ftp 2.4.5, tcllib 1.10 (2006-10-20, bugs 1234831/1581…

    Open ticket 744276ffff · Bug · ftp · filed 2003-05-27

    Full verdict

    PARTLY-FIXED (9.1, 8.6). The reported passive-mode unset ftp(state.data) error was fixed in ftp 2.4.5, tcllib 1.10 (2006-10-20, bugs 1234831/1581453). Active mode still hangs.

    • Module ftp 2.4.14
    • Tested ftpd and the dialog fake server. Passive: Get of a missing file returns 0 / "" at once. Active, when the server opens the data connection at PORT and then answers 550 (tcllib's own ftpd does this): Get hangs forever, and -timeout cannot break it. A leaked data socket also raises a background error (can't read "ftp(Start_Time)") after Close.
    • Cause in the *_sent error branches, ftp(state.data) is only unset in passive mode. In active mode WaitComplete sits in a nested vwait ftp(state.data) that nothing ends. Also, inline Get/Put never record the data socket, so CloseDataConn cannot close it.
    • Prototype prototypes/744276ffff.patch (38 changed lines in ftp.tcl). Tests: ftp-active-error-1.0 (fails before the fix), ftp-active-error-1.1 (Put), ftp-passive-error-1.0. The fix unsets state.data in both modes, has InitDataConn record ftp(DataSock), and makes inline Get call CloseDataConn on failure.
    • Suggested action apply prototype, then close
    • Suggested ticket comment The passive-mode error was fixed in 2006. A related hang remains in active mode when the server connects before answering 550. The patch stops waiting for the data connection on error replies in both modes, closes the leaked socket, and adds tests.
    ⬜664392ffff2003Test failures in uri module for Win32PARTLY-FIXED. Since uri 1.2.7 (tcllib 1.19, kjnash 2016-12/2017-01, e.g. 0b8557b27d), the default is HostAsDriveLetter 0. With that default the W…

    Open ticket 664392ffff · Bug · uri · filed 2003-01-08

    Full verdict

    PARTLY-FIXED. Since uri 1.2.7 (tcllib 1.19, kjnash 2016-12/2017-01, e.g. 0b8557b27d), the default is HostAsDriveLetter 0. With that default the Windows-only file://localhost:/a/b/c join is no longer used, so the canonicalize tests no longer depend on the platform. uri::join scheme file path {c:\windows} still gives file:///c:\windows on trunk. pbrooks is working on a fix in branch pbrooks-ticket-664392 (2026-10-31 .. 2026-11-13, not merged): it converts backslashes on Windows in join, and makes geturl percent-decode the path and convert drive-letter paths to native filenames. That branch bumps uri to 1.2.9.

    • Module uri 1.2.8
    • Tested 9.1 on Linux, split/join/canonicalize of file://localhost/a/b/c round-trips. Read the branch diff.
    • Cause uri.tcl ::uri::JoinFile passes backslashes through.
    • Prototype none (someone is already working on it in a branch)
    • Suggested action keep open. Review and merge pbrooks-ticket-664392 (needs a Windows test run).
    • Suggested ticket comment none needed; a fix is in progress in branch pbrooks-ticket-664392.