Open tcllib tickets: Already fixed or done

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 Outdated, superseded, duplicate and misfiled tickets are listed separately.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.

Already fixed or done 70

⬜8fc5a9642e2026MS-Win,tcl9.1b0,autoproxy package: package require registry version conflictFIXED in Tcl: on Windows Tcl 9.1.1, package require registry gives 1.4 and package versions registry lists only 1.4.

Open ticket 8fc5a9642e · Bug · http :: autoproxy · filed 2026-06-24

Full verdict

FIXED in Tcl: on Windows Tcl 9.1.1, package require registry gives 1.4 and package versions registry lists only 1.4.

  • Module autoproxy 1.8.1 (package require registry 1.0, autoproxy.tcl line 154)
  • Tested read the Tcl trunk sources. NEEDS-WINDOWS for a live check.
  • Cause Tcl 9.1b0rc0 provided registry with the Tcl version number.
  • Prototype none
  • Suggested action close as Fixed (on the Tcl side)
  • Suggested ticket comment Fixed in Tcl by e81499de1b: registry is version 1.4 again in 9.1, so package require registry 1.0 works.
Windows VM, run 1
  • Verdict FIXED in Tcl: on Windows Tcl 9.1.1, package require registry gives 1.4 and package versions registry lists only 1.4.
  • Tested tclsh91.
⬜a6712c3f422024Failing me_tcl.test needs results updateFIXED since tcllib 2.0 (c87d543cef, 2024-09-13)

Open ticket a6712c3f42 · Bug · grammar_me · filed 2024-03-25

Full verdict

FIXED since tcllib 2.0 (c87d543cef, 2024-09-13)

  • Module grammar::me::tcl (grammar_me)
  • Tested me_tcl.test passes 104/104 on 9.1, 9.0 and 8.6.
  • Cause Tcl 9 dropped the "unicode" class from string is. c87d543cef ("Tcl 8/9 adaptions in tests") removed it from the expected message.
  • Prototype none (already fixed)
  • Suggested action close as Fixed
  • Suggested ticket comment Fixed by c87d543cef (in 2.0): the expected error message no longer lists "unicode". The tests pass on 9.0, 9.1 and 8.6.
⬜935ac3b65b2022tcllibc 1.21 sha256 compile fails with MS CFIXED on trunk after 2.0 (cc4d362b78, 2026-08-04, "Integrated pbrooks-2026-08-critcl-fixes"; d69b5f3fd3 removed the conflicting SHA256Init/Update/F…

Open ticket 935ac3b65b · Bug · (unused) · filed 2022-06-10

Full verdict

FIXED on trunk after 2.0 (cc4d362b78, 2026-08-04, "Integrated pbrooks-2026-08-critcl-fixes"; d69b5f3fd3 removed the conflicting SHA256Init/Update/Final declarations from sha256.h). NEEDS-WINDOWS to confirm.

  • Module sha256c (sha256 1.0.6)
  • Tested code inspection only. Current sha256.h no longer declares the functions, so only the stubs header (sha256cDecls.h) declares them. No critcl/MSVC here.
  • Cause duplicate prototypes with different linkage in sha256.h vs the generated stubs declarations
  • Prototype none
  • Suggested action confirm with an MSVC build of trunk, then close as Fixed. The old leaves sha256-vc-fix (3d04580503) and apn-vc-fixes (3b2ac7132c) can then be closed.
  • Suggested ticket comment The conflicting declarations were removed from sha256.h on trunk in cc4d362b78. Please confirm with an MSVC build of trunk. Branch sha256-vc-fix is then obsolete.
⬜da67f546e22020Tcllib vs Tcl 8.7aX - Testsuite: 6 failures across 5 modulesFIXED (no failures on current Tcl)

Open ticket da67f546e2 · Bug · Build System & Testing · filed 2020-03-07

Full verdict

FIXED (no failures on current Tcl)

  • Module cmdline, fileutil, grammar_me, snit, tar tests
  • Tested sak test run on cmdline fileutil grammar_me snit tar with 9.1, 9.0 and 8.6: 2025 passed, 0 failed out of 2103 (the rest skipped) on each
  • Cause 8.7a-era differences, adapted since
  • Prototype none
  • Suggested action close as Fixed
  • Suggested ticket comment These five modules pass on 8.6, 9.0 and 9.1 with no failures.
⬜7733d51d632019error building crc32 tcllibc code on WinFIXED since tcllib 2.0 (9a3a6d56be, 2024-01-11). The critcl code moved out of crc32.tcl into crc32c.tcl. In 1.20 crc32.tcl defined Crc32_c itself…

Open ticket 7733d51d63 · Bug · crc · filed 2019-12-24

Full verdict

FIXED since tcllib 2.0 (9a3a6d56be, 2024-01-11). The critcl code moved out of crc32.tcl into crc32c.tcl. In 1.20 crc32.tcl defined Crc32_c itself and was also pulled in again via yencode's package require crc32, hence "redefinition of tcl__crc_Crc32_c".

  • Module crc32 1.3.4 / crcc
  • Tested code inspection of tcllib-1-20, tcllib-1-21 and trunk. No critcl here, so no build was tried.
  • Prototype none (fixed)
  • Suggested action close as Fixed
  • Suggested ticket comment Since tcllib 2.0 the critcl implementation lives only in crc32c.tcl, so crc32.tcl no longer carries C code and the double definition cannot happen.
⬜c824f2425b2019Default configure finds Tcl 9, failsFIXED since tcllib 2.0 (758418619e, 2023-11-14, Rolf's Tcl 9 work integrated; installer now package require Tcl 8.5 9)

Open ticket c824f2425b · Bug · (unused) · filed 2019-12-02

Full verdict

FIXED since tcllib 2.0 (758418619e, 2023-11-14, Rolf's Tcl 9 work integrated; installer now package require Tcl 8.5 9)

  • Module installer.tcl / configure (trunk)
  • Tested VPATH configure with Tcl 9 first on PATH (it picked /bin/tclsh9.0), then make install-libraries: completed and generated tcllib2.0/pkgIndex.tcl
  • Cause installer.tcl required Tcl 8.2 (exact major)
  • Prototype none
  • Suggested action close as Fixed
  • Suggested ticket comment Tcllib 2.0 supports Tcl 9, and the installer accepts 8.5–9. configure/make install with a Tcl 9 tclsh works on trunk.
⏳4aeded587b2018https://core.tcl.tk/tcllib/doc/trunk/embedded/www/tcllib/files/modules/asn/asn.htmlDONE on trunk (8c56f46b22, 2026-01-30, mark: "added an example to the asn package's docs"; after tcllib 2.0, not in a release yet). asn.man's Exa…

Open ticket 4aeded587b · Bug · asn · filed 2018-06-29 · Pending (resolved, waiting for confirmation)

⏳ Since triage (updated 2026-11-16)

Status now: Pending

    Full verdict

    DONE on trunk (8c56f46b22, 2026-01-30, mark: "added an example to the asn package's docs"; after tcllib 2.0, not in a release yet). asn.man's Examples section now has a full encode/decode example.

    • Module asn 0.8.5 (asn.man)
    • Tested read asn.man on trunk
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed (trunk, next release)
    • Suggested ticket comment The asn manpage now has an inline encode/decode example (8c56f46b22). Closing.
    ⬜a7fb3a643c2017math::exact relies on bad Tcl resolution behaviour revoked by TIP 278 - Tcl 9FIXED since tcllib 1.19 (fb859d7803, 2018-01-24)

    Open ticket a7fb3a643c · Bug · math · filed 2017-10-31

    Full verdict

    FIXED since tcllib 1.19 (fb859d7803, 2018-01-24)

    • Module math, math::exact
    • Tested exact.test on 9.1, 9.0, 8.6: 198/198 pass. No relative $math::... variable references are left. The remaining relative math::exact::... names are command names, which TIP 278 did not change.
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed by fb859d7803. math::exact tests pass on Tcl 9.0 and 9.1.
    ⬜43672018172017yaml relies on bad Tcl resolution behaviour revoked by TIP 278 - Tcl 9FIXED since tcllib 1.19 (d50864197b, 2018-01-24), plus the Tcl 9 port 22c749a4ef (2023, in tcllib 2.0)

    Open ticket 4367201817 · Bug · yaml · filed 2017-10-31

    Full verdict

    FIXED since tcllib 1.19 (d50864197b, 2018-01-24), plus the Tcl 9 port 22c749a4ef (2023, in tcllib 2.0)

    • Module yaml 0.4.2
    • Tested yaml module tests 156/156 on 9.1 and 8.6; no $yaml::-style relative variable references remain.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Relative yaml:: variable references were removed in d50864197b, and yaml passes its tests on Tcl 9.0/9.1. Closing.
    ⬜a12bff914d2017add smtp.tcl support direct TLS connection with backward compatibleDONE (657782a889 / 9073bd414a, 2018, ticket 83d674b2dd; option -tlsimport)

    Open ticket a12bff914d · Patch · smtp · filed 2017-10-27

    Full verdict

    DONE (657782a889 / 9073bd414a, 2018, ticket 83d674b2dd; option -tlsimport)

    • Module mime / smtp 1.5.2
    • Tested code and docs. -tlsimport 1 does tls::import right after the socket opens (native TLS, e.g. port 465). The attached patch artifact is not in the repository.
    • Cause -
    • Prototype none
    • Suggested action close as Fixed (superseded by -tlsimport)
    • Suggested ticket comment Direct TLS is supported via smtp::sendmessage -tlsimport 1 (ticket 83d674b2dd, tcllib 1.20). Closing.
    ⬜0bd02c1bf82016websocket: sec-websocket-protocol is optional but checked in websocket::testFIXED since tcllib 2.0 (25005159c5, merged a500ff9409, 2022-10-10). DUPLICATE of e23279c7d1 and 0dd2a66f08.

    Open ticket 0bd02c1bf8 · Bug · websocket · filed 2016-11-20

    Full verdict

    FIXED since tcllib 2.0 (25005159c5, merged a500ff9409, 2022-10-10). DUPLICATE of e23279c7d1 and 0dd2a66f08.

    • Module websocket 1.6
    • Tested 9.1, websocket::test without the protocol header.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed in tcllib 2.0 (25005159c5, from ticket 0dd2a66f08): validate now returns an empty protocol list when the header is absent.
    ⬜c4947e1f352016Write manpages for httpDONE (httpd.man added in 5dd3f483bf/df153cb3ea, 2018-02, tcllib 1.19; now generated from the sources by build/build.tcl)

    Open ticket c4947e1f35 · Bug · A Category Is Missing · filed 2016-11-14

    Full verdict

    DONE (httpd.man added in 5dd3f483bf/df153cb3ea, 2018-02, tcllib 1.19; now generated from the sources by build/build.tcl)

    • Module httpd 4.3.x
    • Tested modules/httpd/httpd.man exists and is in the generated docs
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment httpd has had a manpage since tcllib 1.19 (2018). It is now generated from the sources. Closing.
    ⬜3bb067b6672016package load failureFIXED since tcllib 2.0 (bb98482e6f, 2024-09-13, "Tcl 9 portability fixes")

    Open ticket 3bb067b667 · Bug · snit · filed 2016-04-26

    Full verdict

    FIXED since tcllib 2.0 (bb98482e6f, 2024-09-13, "Tcl 9 portability fixes")

    • Module snit 2.3.4 / 1.4.3
    • Tested set library x; source snit2.tcl on 9.1/8.6 now works, and ::library is untouched. Old artifacts up to 2023 still had set library ... inside namespace eval ::snit::, and that mechanism fails on 8.6 (info exists ::snit::library → 0).
    • Cause set instead of variable in snit.tcl/snit2.tcl; now variable library.
    • Prototype none (fixed)
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed in tcllib 2.0 (snit 2.3.4/1.4.3, check-in bb98482e6f): the loaders now use variable library.
    ⬜4842c9ce0b2016::uuid::uuid generate: couldn't execute "ifconfig": no such file or directoryFIXED since tcllib 1.20 (d5e2fcf04b, 2018-05-31: use /dev/urandom when it exists instead of nettool)

    Open ticket 4842c9ce0b · Bug · uuid · filed 2016-02-18

    Full verdict

    FIXED since tcllib 1.20 (d5e2fcf04b, 2018-05-31: use /dev/urandom when it exists instead of nettool)

    • Module uuid 1.0.9
    • Tested env PATH=/bin tclsh with uuid::uuid generate works on 9.1 and 8.6. Also, nettool cannot be loaded at all right now (see the pkgIndex typo above), so uuid never reaches nettool.
    • Cause n/a. The remaining exec ifconfig in nettool (build/platform_unix_linux.tcl) is covered by tickets d879576438 and 9a21637273.
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Since 2018, uuid uses /dev/urandom when present and no longer calls nettool/ifconfig on Unix. Closing; the nettool ifconfig dependency is tracked in d879576438.
    ⬜e23279c7d12015Websocket: empty protocol list throws errorFIXED since tcllib 2.0 (25005159c5, ticket 0dd2a66f08; merged in a500ff9409, 2022-10-10). validate now always returns a protocols key. Also a D…

    Open ticket e23279c7d1 · Bug · websocket · filed 2015-06-15

    Full verdict

    FIXED since tcllib 2.0 (25005159c5, ticket 0dd2a66f08; merged in a500ff9409, 2022-10-10). validate now always returns a protocols key. Also a DUPLICATE of 0bd02c1bf8.

    • Module websocket 1.6
    • Tested 9.1, websocket::test with no Sec-WebSocket-Protocol header returns 1.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed in tcllib 2.0 (25005159c5): a missing Sec-WebSocket-Protocol header now gives an empty protocol list, and any handler matches.
    ⬜c3478ceae22015New json to huddle parserDONE (02ec11eeba, 2015-11-24, Tcllib 1.18). It brought json2huddle.tcl, the number/boolean/null types, combine_relaxed, the isContainer setti…

    Open ticket c3478ceae2 · Bug · huddle · filed 2015-05-04

    Full verdict

    DONE (02ec11eeba, 2015-11-24, Tcllib 1.18). It brought json2huddle.tcl, the number/boolean/null types, combine_relaxed, the isContainer setting and the matching yaml changes.

    • Module yaml/huddle::json 0.2 (json2huddle.tcl), huddle 0.5
    • Tested the files and features are present on trunk. json2huddle.test passes as part of the yaml module tests (9.1, 8.6).
    • Cause -
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Merged with the huddle 0.2 / yaml 0.3.8 work in Tcllib 1.18 (02ec11eeba), including huddle::json.
    ⬜34cf417ffd2015comm-6.0 failsFIXED since tcllib 1.19 (2c24d9d9c1, 2018-01-08: comm-6.0 now uses -ssl2 0 -ssl3 0 -tls1 1)

    Open ticket 34cf417ffd · Bug · comm · filed 2015-04-30

    Full verdict

    FIXED since tcllib 1.19 (2c24d9d9c1, 2018-01-08: comm-6.0 now uses -ssl2 0 -ssl3 0 -tls1 1)

    • Module comm 4.7.3
    • Tested CANNOT-TEST the tls test itself (no tls package installed for 9.1/9.0/8.6, test skipped); checked comm.test via annotate
    • Cause test asked for SSLv2/v3 only, refused by modern OpenSSL
    • Prototype none (fixed)
    • Suggested action close as Fixed
    • Suggested ticket comment comm-6.0 was changed to use TLS 1 instead of SSLv2/v3 in tcllib 1.19.
    ⬜9cbe339d782015pop3-7.0 failsFIXED since tcllib 1.19 (c49c11b395 "Do not use ssl2 or ssl3 in tests", patthoyts, 2017-10-27, merged 048189d3a6). The test server was SSLv3-only, …

    Open ticket 9cbe339d78 · Bug · pop3 · filed 2015-04-30

    Full verdict

    FIXED since tcllib 1.19 (c49c11b395 "Do not use ssl2 or ssl3 in tests", patthoyts, 2017-10-27, merged 048189d3a6). The test server was SSLv3-only, which modern OpenSSL rejects.

    • Module pop3 1.11
    • Tested CANNOT run pop3-7.0 here (no tcltls; the test is skipped on 9.1/8.6). Caveat: the test certificates in modules/devtools are 1024-bit RSA, which OpenSSL 3 at the default security level 2 may also reject. Whoever has tcltls should re-run it.
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment The test used an SSLv3-only server; since 1.19 it uses TLS 1.x (c49c11b395). Please reopen if it still fails with a current tcltls.
    ⬜b5092d56232015(title is an e-mail address; ooutil tests fail)FIXED since tcllib 1.18 (f705e6446e, 2015-05-21, per aku's comment)

    Open ticket b5092d5623 · Patch · Website · filed 2015-04-30

    Full verdict

    FIXED since tcllib 1.18 (f705e6446e, 2015-05-21, per aku's comment)

    • Module oo::util 1.2.3
    • Tested sak test run modules/ooutil on 9.1 and 8.6: 4/4 pass.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed. The title contains an email address and there are three later anonymous spam comments with personal data; consider retitling to "ooutil tests: can't find package T" and hiding the spam comments.
    • Suggested ticket comment Fixed by f705e6446e in 2015; ooutil tests pass on 8.6 and 9.x. Closing.
    ⬜a753cade832015New improved huddle versionDONE (02ec11eeba "Merged huddle/yaml work - Updated to huddle 0.2, yaml 0.3.8", 2015-11-24. It was developed on branch huddle-a753cade83 and late…

    Open ticket a753cade83 · Bug · huddle · filed 2015-04-23

    Full verdict

    DONE (02ec11eeba "Merged huddle/yaml work - Updated to huddle 0.2, yaml 0.3.8", 2015-11-24. It was developed on branch huddle-a753cade83 and later updated in 5a516799b4, 2015-12-10)

    • Module yaml/huddle 0.5
    • Tested history review (fossil finfo modules/yaml/huddle.tcl)
    • Cause -
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Integrated as huddle 0.2 in Tcllib 1.18 via branch huddle-a753cade83 (02ec11eeba).
    ⬜58eb8d4e842015Boolean type and a bugDONE (02ec11eeba, 2015-11-24, Tcllib 1.18). It added huddle true/false/boolean (tag b) with JSON output true/false. huddle create now use…

    Open ticket 58eb8d4e84 · Bug · huddle · filed 2015-04-16

    Full verdict

    DONE (02ec11eeba, 2015-11-24, Tcllib 1.18). It added huddle true/false/boolean (tag b) with JSON output true/false. huddle create now uses dict semantics for duplicate keys.

    • Module yaml/huddle 0.5
    • Tested huddle create a 1 a 2 gives HUDDLE {D {a {s 2}}} on trunk (1.17 gave {a {s 1} a {s 2}}). huddle jsondump [huddle create a [huddle boolean true]] gives {"a":true}. Checked on 9.1 and 8.6.
    • Cause -
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Both points were done in the huddle 0.2 rewrite (Tcllib 1.18): there is a boolean type, and create keeps only the last value of a duplicate key.
    ⬜071dbedcc82015Waiting on testcases for fix to mishandled token "string" and new feature "casesensitivestrings"patchDONE (cd1a3ac12d, tcllib 1.17). Only the test cases were missing, and they are now written.

    Open ticket 071dbedcc8 · Bug · bibtex · filed 2015-03-19

    Full verdict

    DONE (cd1a3ac12d, tcllib 1.17). Only the test cases were missing, and they are now written.

    • Module bibtex 0.8
    • Tested new tests bibtex-6.0..6.3 pass on 9.1 and 8.6. bibtex-6.0 fails when the ^ anchor is removed.
    • Cause n/a
    • Prototype prototypes/071dbedcc8.patch (32 lines, tests only: bibtex-6.0 "string" inside a field, 6.1 case-insensitive default, 6.2 -casesensitivestrings 1, 6.3 quoted macro value)
    • Suggested action apply the test patch and close as Fixed
    • Suggested ticket comment Added tests for the "string"-inside-field fix and for -casesensitivestrings. Closing.
    ⬜c601e3dc6c2014String is double does not conform to JSON specificationFIXED since tcllib 1.18 (02ec11eeba, 2015-11-24). This is the same issue as 3315620fff.

    Open ticket c601e3dc6c · Patch · huddle · filed 2014-09-04

    Full verdict

    FIXED since tcllib 1.18 (02ec11eeba, 2015-11-24). This is the same issue as 3315620fff.

    • Module yaml/huddle 0.5
    • Tested huddle jsondump [huddle create a b c 00.34] gives {"a":"b","c":"00.34"} on 9.1 and 8.6. The attached patch (artifact 73c511b240) is not in the local repo, so it could not be compared, but its stated result now matches the trunk output.
    • Cause -
    • Prototype none
    • Suggested action close as Fixed (or as a duplicate of 3315620fff)
    • Suggested ticket comment Since huddle 0.2 (Tcllib 1.18) string values are always quoted, so 00.34 is dumped as "00.34". Numbers must be created with huddle number.
    ⬜a01e30ff402014websocket documentation still lacking in placesDONE (a500ff9409, 2022-10-10, tcllib 2.0) for the example given: the error type is now documented in websocket.man together with connect, disconn…

    Open ticket a01e30ff40 · Bug · websocket · filed 2014-08-12

    Full verdict

    DONE (a500ff9409, 2022-10-10, tcllib 2.0) for the example given: the error type is now documented in websocket.man together with connect, disconnect, close and timeout.

    • Module websocket 1.6
    • Tested read websocket.man.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed. Remaining gaps, such as the test/upgrade context (577ed0d10b), are tracked separately.
    • Suggested ticket comment The error type and the other event types are documented now (since a500ff9409). Closing; specific gaps can get their own tickets.
    ⬜fb97352ad92014chase compiler warningsDONE. Unused-variable/initialization parts went in with 5d35be9b3a (2014-08-11). The pointer-cast parts were done differently: `(void*)(uintptr_t…

    Open ticket fb97352ad9 · Patch · pt (parsetools) · filed 2014-08-07

    Full verdict

    DONE. Unused-variable/initialization parts went in with 5d35be9b3a (2014-08-11). The pointer-cast parts were done differently: (void*)(uintptr_t) with <stdint.h> in 4bc853f186 (2020-05-15, "Fixed all the warnings reported on Linux"), then 022eea6e5b (2024, Tcl 9) and 0b25980a00 (2026-10-02, Windows warnings).

    • Module pt rde_critcl (C accelerator)
    • Tested compared the attached patch chunks with the current rde_critcl/param.c, p.c and tc.c. Every cast site now uses (u)intptr_t.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Unused variables were fixed in 5d35be9b3a. The cast warnings were later fixed with (u)intptr_t casts in 4bc853f186 and the 2026 warning cleanup 0b25980a00. Closing.
    ⬜3536709fff2012huddle get does not accept non-scalar dict keysFIXED since tcllib 1.18 (huddle 0.2 rewrite, 02ec11eeba, 2015-11-24)

    Open ticket 3536709fff · Patch · yaml · filed 2012-06-20

    Full verdict

    FIXED since tcllib 1.18 (huddle 0.2 rewrite, 02ec11eeba, 2015-11-24)

    • Module huddle 0.5
    • Tested huddle get [huddle create {a a} b] {a a} works on 9.1 and 8.6 trunk (also get_stripped, exists, nested paths). The tcllib-1-17 huddle.tcl (0.1.5) fails with key "{a a}" not known in dictionary.
    • Cause n/a (old _key_reflexive is gone)
    • Prototype none (no regression test exists for it; the reporter's test case could be added)
    • Suggested action close as Fixed
    • Suggested ticket comment Since the huddle 0.2 rewrite (tcllib 1.18), huddle get $h {a a} works. Closing as fixed.
    ⬜3513149fff2012exra bracket in nvlist command causes corrupt return valueFIXED since tcllib 1.15 (2a9631a53b, 2012-03-30, ncgi 1.3.3). The ticket resolution is already "Fixed" but the status is still Open.

    Open ticket 3513149fff · Bug · ncgi · filed 2012-03-30

    Full verdict

    FIXED since tcllib 1.15 (2a9631a53b, 2012-03-30, ncgi 1.3.3). The ticket resolution is already "Fixed" but the status is still Open.

    • Module ncgi 1.4.6
    • Tested code reading; ncgi::reset "=x"; ncgi::nvlist gives anonymous x
    • Cause n/a
    • Prototype none
    • Suggested action close (status Closed, resolution already Fixed)
    • Suggested ticket comment Fixed in ncgi 1.3.3 (tcllib 1.15). Closing.
    ⬜3315620fff2011Invalid JSON with values that begin with 0FIXED since tcllib 1.18 (02ec11eeba, 2015-11-24). Plain values are huddle strings and are always quoted; numbers need an explicit huddle number.

    Open ticket 3315620fff · Bug · huddle · filed 2011-06-13

    Full verdict

    FIXED since tcllib 1.18 (02ec11eeba, 2015-11-24). Plain values are huddle strings and are always quoted; numbers need an explicit huddle number.

    • Module yaml/huddle 0.5
    • Tested huddle jsondump [huddle create key 000000] gives {"key":"000000"} on 9.1 and 8.6 (1.17 gave 000000).
    • Cause -
    • Prototype none. Residual issue: huddle number / huddle compile number validate with string is double, so huddle number 007 (also 0x10, Inf, " 12") still dumps invalid JSON. That is a different and narrower issue, left as is because yaml may rely on such tokens.
    • Suggested action close as Fixed
    • Suggested ticket comment Since huddle 0.2 (Tcllib 1.18) strings are always quoted in jsondump; numbers must be created explicitly with huddle number.
    ⬜3065746fff2010formatted log::log for performenceDONE (log::logsubst, ea802e332b, 2018-05-18, log 1.4, tcllib 1.20. Implemented by the same reporter as RFE 19607f927b)

    Open ticket 3065746fff · RFE · log · filed 2010-09-14

    Full verdict

    DONE (log::logsubst, ea802e332b, 2018-05-18, log 1.4, tcllib 1.20. Implemented by the same reporter as RFE 19607f927b)

    • Module log 1.5
    • Tested log.man documents ::log::logsubst level msg with lazy substitution. Its tests check that suppressed levels do not substitute.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed (superseded by 19607f927b).
    • Suggested ticket comment Implemented as log::logsubst (subst-based, as suggested in the comments) via 19607f927b, log 1.4, tcllib 1.20.
    ⬜3042811fff2010huddle bug in dicts with complex keysFIXED since tcllib 1.18 (02ec11eeba, 2015-11-24, huddle 0.2 rewrite)

    Open ticket 3042811fff · Bug · huddle · filed 2010-08-11

    Full verdict

    FIXED since tcllib 1.18 (02ec11eeba, 2015-11-24, huddle 0.2 rewrite)

    • Module yaml/huddle 0.5
    • Tested huddle jsondump [huddle create {a b} b c d] gives {"a b":"b","c":"d"} on trunk with 9.1 and 8.6, and on the tcllib-1-18 huddle.tcl. The tcllib-1-17 huddle.tcl still fails with key "{a b}" not known in dictionary.
    • Cause -
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed by the huddle 0.2 rewrite in Tcllib 1.18: keys with spaces now work in create/get/jsondump.
    ⬜2979481fff2010string macro parsing is buggyFIXED since tcllib 1.17 (cd1a3ac12d, 2015-04-21; branch bibtex-071dbedcc8)

    Open ticket 2979481fff · Bug · bibtex · filed 2010-03-30

    Full verdict

    FIXED since tcllib 1.17 (cd1a3ac12d, 2015-04-21; branch bibtex-071dbedcc8)

    • Module bibtex 0.8
    • Tested title = {A string {theory} test} now parses as a normal record on 9.1 and 8.6
    • Cause the regexp lacked a ^ anchor. ParseRecords now uses {^\s*string[^\{]*\{(.*)\}[^\}]*}.
    • Prototype none. The regression test is in prototypes/071dbedcc8.patch (bibtex-6.0).
    • Suggested action close as Fixed (duplicate of 071dbedcc8)
    • Suggested ticket comment Fixed in tcllib 1.17 by the patch from 071dbedcc8. The string regexp is now anchored at the start of the entry.
    ⬜2845453fff2009MessageReceiver fails to read sequence byte on empty socketFIXED since tcllib 1.19 (8b8f10e2bb, 2017-11-01, ticket c247ed5db4, merged 0826350f8d 2018-01-24)

    Open ticket 2845453fff · Bug · ldap · filed 2009-08-27

    Full verdict

    FIXED since tcllib 1.19 (8b8f10e2bb, 2017-11-01, ticket c247ed5db4, merged 0826350f8d 2018-01-24)

    • Module ldap 1.10.2
    • Tested code review. ldap::MessageReceiver now has a partial branch for the sequence byte that just returns and waits for the next fileevent. It is present in tcllib-1-19 (ldap 1.9.2). CANNOT-TEST the original LDAPS server situation.
    • Cause the old code sent a "partial" (zero-byte) non-blocking read to the default branch, which reported a protocol error.
    • Prototype none
    • Suggested action close as Fixed (duplicate of c247ed5db4).
    • Suggested ticket comment Fixed by 8b8f10e2bb for ticket c247ed5db4 (ldap 1.9.2, tcllib 1.19): a partial read of the sequence byte now waits for the next readable event instead of failing.
    ⬜2680202fff2009memory leakFIXED since tcllib 1.12 (2ba9a8f7a5, 2009-05-06, "md5c command was leaking a Tcl_Obj on each call")

    Open ticket 2680202fff · Bug · md5 · filed 2009-03-10

    Full verdict

    FIXED since tcllib 1.12 (2ba9a8f7a5, 2009-05-06, "md5c command was leaking a Tcl_Obj on each call")

    • Module md5 2.0.9 / md5c 0.12 (critcl)
    • Tested no critcl available here, so I built md5c.tcl's C code by hand as a stubs extension (9.1, 8.6). Then ran 4×(20000 md5::md5 string + 300 md5 -file on 100 KB) with accel(critcl)=1: RSS flat (9212→9216 KB on 9.1, 6872→6876 KB on 8.6)
    • Cause extra Tcl_IncrRefCount in md5c (reporter used tcllibc 0.3.4 from 1.11)
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment The md5c refcount leak was fixed in 2ba9a8f7a5 (tcllib 1.12). The current md5c shows no growth over 80000 calls on Tcl 8.6 and 9.1.
    ⬜2444762fff2008Mozilla 3.x breaks ncgi::nvlistFIXED since tcllib 1.9 (bd21455bc5, 2006-07-02, ncgi 1.3.1), the same fix as 1116123. The error message quoted ("dont know how to handle ...") come…

    Open ticket 2444762fff · Bug · ncgi · filed 2008-12-18

    Full verdict

    FIXED since tcllib 1.9 (bd21455bc5, 2006-07-02, ncgi 1.3.1), the same fix as 1116123. The error message quoted ("dont know how to handle ...") comes from tclhttpd, not ncgi, so the reporter probably had an older ncgi bundled with tclhttpd.

    • Module ncgi 1.4.6
    • Tested urlencoded data with Content-Type: application/x-www-form-urlencoded; charset=UTF-8 parses fine on 9.1 and 8.6
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed (or as a duplicate of 1116123)
    • Suggested ticket comment ncgi 1.3.1 and later accept "; charset=..." after the urlencoded content type. The quoted message comes from tclhttpd.
    ⬜2414880fff2008fix installer.tcl to look for and handle tcllibcDONE (4b2ffb42f2, 2019-08-29, ticket 138d56dbaf, tcllib 1.20)

    Open ticket 2414880fff · Bug · Build System & Testing · filed 2008-12-10

    Full verdict

    DONE (4b2ffb42f2, 2019-08-29, ticket 138d56dbaf, tcllib 1.20)

    • Module installer.tcl / support/installation
    • Tested code reading. support/installation/modules.tcl registers tcllibc when modules/tcllibc exists. make install-binaries (sak.tcl critcl) builds and installs it.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Done in 1.20 (4b2ffb42f2): the installer installs a tcllibc it finds, and make install-binaries builds and installs it via critcl.
    ⬜2225148fff2008struct::list test suite problems against tcl 8.6 cvs headFIXED since tcllib 1.12 (3846dac25c, 2009-09-24)

    Open ticket 2225148fff · Bug · struct :: list · filed 2008-11-05

    Full verdict

    FIXED since tcllib 1.12 (3846dac25c, 2009-09-24)

    • Module struct::list 1.8.x (modules/struct/list.tcl)
    • Tested sak test run modules/struct on 9.1 and 8.6: 0 failures. assign-4.4, repeat-4.2 and repeat-4.4 now carry the tcl8.5only constraint, and repeat-4.5 uses a per-version result.
    • Cause TIP 323 changed the core lassign/lrepeat that struct::list aliases on 8.5+. The tests were adjusted in 3846dac25c under the duplicate SF bug 2557046 (Closed/Fixed).
    • Prototype none (already fixed)
    • Suggested action close as Fixed (duplicate of 2557046fff)
    • Suggested ticket comment Fixed by 3846dac25c (Bug 2557046, released in Tcllib 1.12). The tests that check the TIP 323 behaviour are now limited to Tcl 8.5 or get per-version results. The struct tests pass on Tcl 8.6 and 9.1.
    ⬜2022905fff200874 test failures with Tcl HEAD (wrong # args)FIXED (the testsuite was adapted for 8.6's wrong#args messages long ago; tcllib trunk tests run clean on 8.6 and 9.x)

    Open ticket 2022905fff · Bug · Build System & Testing · filed 2008-07-20

    Full verdict

    FIXED (the testsuite was adapted for 8.6's wrong#args messages long ago; tcllib trunk tests run clean on 8.6 and 9.x)

    • Module whole testsuite
    • Tested full sak test run on trunk. 9.1: 26708 passed, 13 failed out of 27126. 8.6: 27038 passed, 4 failed out of 27445. None of the failures is a wrong#args message. The 9.1-only failures are crc (7, new binary "expected byte sequence" message), hook (2, changed -errorstack contents) and doctools2base msgcat-2.0, which fails on both versions; websocket (3) also fails on both.
    • Cause 2008-era 8.6a message reform. The testutilities tcltest::wrongNumArgs/tcl-version helpers now handle it.
    • Prototype none
    • Suggested action close as Out of Date
    • Suggested ticket comment The 2008 failures are long gone. The current suite passes on Tcl 8.6 and 9.x without wrong#args failures.
    ⬜2002489fff2008tree-tcl test 4.4.17 in test suite - error msg differenceFIXED since tcllib 1.11 (stack traces moved into per-version files, ChangeLog 2008-09-02)

    Open ticket 2002489fff · Bug · treeql · filed 2008-06-25

    Full verdict

    FIXED since tcllib 1.11 (stack traces moved into per-version files, ChangeLog 2008-09-02)

    • Module struct::tree 2.1.3
    • Tested sak test run modules/struct on 9.1, 8.6 and 8.5: 0 failures. tree-tcl-4.4.17 and 4.4.17.0-6 pass.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment tree-tcl-4.4.17 passes on Tcl 8.5, 8.6, 9.0 and 9.1 since the 2008 rework of the expected stack traces. Closing.
    ⬜2001950fff2008encoding of ini fileDONE (1a2e9c5b7b, 2014-07-21, ticket c4b8162da5; inifile 0.3, in tcllib 1.17)

    Open ticket 2001950fff · RFE · inifile · filed 2008-06-24

    Full verdict

    DONE (1a2e9c5b7b, 2014-07-21, ticket c4b8162da5; inifile 0.3, in tcllib 1.17)

    • Module inifile 0.3.3
    • Tested code read. ini::open ?-encoding E? file ?mode? exists. Tests inifile-4.x cover it. b09616777a (2018) added encoding handling to commit/revert.
    • Cause n/a
    • Prototype none (implemented)
    • Suggested action close as Fixed (duplicate of c4b8162da5)
    • Suggested ticket comment Implemented by 1a2e9c5b7b (ticket c4b8162da5): ini::open accepts -encoding since inifile 0.3 (Tcllib 1.17).
    ⬜1928856fff2008fsCmd list - no socket as parameter?FIXED since tcllib 1.14 (ftpd 1.2.6, [Bug 3357765] passive-mode fix with PasvCheckAndWait, ChangeLog 2011-08-09)

    Open ticket 1928856fff · Bug · ftpd · filed 2008-03-29

    Full verdict

    FIXED since tcllib 1.14 (ftpd 1.2.6, [Bug 3357765] passive-mode fix with PasvCheckAndWait, ChangeLog 2011-08-09)

    • Module ftpd 1.4.1
    • Tested ftp::NList against ftpd, passive and active (9.1, 8.6). The listing arrives over the data socket, so dlist receives the channel. In 2008, List called Fs dlist right after PASV, before the client had connected, so the channel was still "". Remaining corner case: a LIST/NLST sent without any PORT/PASV still passes "" to fsCmd instead of replying 425.
    • Cause was ftpd.tcl ::ftpd::List not waiting for the passive connection
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed by the passive-mode rework in ftpd 1.2.6 (tcllib 1.14). List now waits for the data connection and passes its channel to fsCmd dlist.
    ⬜1897841fff2008nameprep test 5 failspatchFIXED (by Tcl: passes on 9.1, 9.0, 8.6 tip; still fails on 8.5)

    Open ticket 1897841fff · Bug · stringprep · filed 2008-02-20

    Full verdict

    FIXED (by Tcl: passes on 9.1, 9.0, 8.6 tip; still fails on 8.5)

    • Module stringprep 1.0.3
    • Tested stringprep.test -constraints knownBug passes 25/25 on 9.1, 9.0 and 8.6. On 8.5 nameprep-5 still fails: no non-BMP characters (U+1D7BB).
    • Cause Tcl versions without full Unicode could not represent U+1D7BB. The test is currently always skipped via knownBug.
    • Prototype prototypes/1897841fff.patch (39 lines, test-only). It replaces knownBug with a fullUnicode constraint (true when encoding convertfrom utf-8 yields U+1D7BB), so nameprep-5 runs on 8.6+/9.x and is skipped on 8.5. Passes on 9.1/8.6, skipped on 8.5.
    • Suggested action apply prototype, close as Fixed
    • Suggested ticket comment Passes now on Tcl 8.6 and 9.x, which handle U+1D7BB; only 8.5 lacks it. Proposed test change: run nameprep-5 under a fullUnicode constraint instead of knownBug.
    ⬜1843712fff2007Option to set client addressDONE (ChangeLog 2007-01-25, FR 1614860)

    Open ticket 1843712fff · RFE · smtp · filed 2007-12-03

    Full verdict

    DONE (ChangeLog 2007-01-25, FR 1614860)

    • Module mime / smtp 1.5.2
    • Tested -client is in sendmessage and documented in smtp.man.
    • Cause -
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment smtp::sendmessage -client exists and is documented. It was added in January 2007 (FR 1614860). Closing.
    ⬜1826418fff2007Various syntax errorsFIXED since tcllib 1.11 (2008-03-15, per aku: spf, term, grammar_fa/fa_op, dns etc.; ChangeLogs cite SF 1826418). The other reports were artifacts …

    Open ticket 1826418fff · Bug · None · filed 2007-11-05

    Full verdict

    FIXED since tcllib 1.11 (2008-03-15, per aku: spf, term, grammar_fa/fa_op, dns etc.; ChangeLogs cite SF 1826418). The other reports were artifacts of an outdated procheck.

    • Module various
    • Tested ChangeLog greps (term, grammar_fa, dns reference 1826418). Trunk is checked with nagelfar (##nagelfar pragmas), and sak validate exists.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed.
    • Suggested ticket comment The real errors were fixed in 2008. The rest were false positives of an old procheck. Closing.
    ⬜1815519fff2007fileutil::find does not return symlinksFIXED since tcllib 1.11 (605bf758ee, 2007-10-24: "Changed handling of broken symbolic links ... They are now recognized and returned", fileutil 1.1…

    Open ticket 1815519fff · Bug · fileutil · filed 2007-10-18

    Full verdict

    FIXED since tcllib 1.11 (605bf758ee, 2007-10-24: "Changed handling of broken symbolic links ... They are now recognized and returned", fileutil 1.13.4; tests find-1.10.x)

    • Module fileutil 1.16.4, fileutil::traverse 0.7
    • Tested a directory with a file, a good link, a dangling link and a dir link. fileutil::find and fileutil::traverse return all of them on 9.1, 8.6, 8.5.
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed in 2007 (fileutil 1.13.4, Tcllib 1.11). Broken symlinks are now returned by find and traverse (tests find-1.10.*).
    ⬜1720896fff2007test suite enhancement - bug report filesDONE (mostly), otherwise OUTDATED

    Open ticket 1720896fff · RFE · general request · filed 2007-05-17

    Full verdict

    DONE (mostly), otherwise OUTDATED

    • Module sak
    • Tested sak test run -l STEM writes STEM.log, STEM.summary, STEM.failures, STEM.skipped and STEM.none. The part about submitting to sf.net no longer applies (tickets are in fossil).
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed/Out of date
    • Suggested ticket comment sak test run -l STEM now writes the detailed log and a separate failures list, which can be attached to a ticket. Closing.
    ⬜1695120fff2007doctools - add diagram supportDONE (doctools 1.4.7, 2010-06-08, added the [image name ?label?] markup. dtplite maps symbolic names to png/jpeg/pic/txt variants. Tcllib's own d…

    Open ticket 1695120fff · RFE · doctools · filed 2007-04-05

    Full verdict

    DONE (doctools 1.4.7, 2010-06-08, added the [image name ?label?] markup. dtplite maps symbolic names to png/jpeg/pic/txt variants. Tcllib's own docs, e.g. modules/pt/include/*.dia, keep diagrams as tklib diagram sources that are rendered to images.)

    • Module doctools 1.6.1
    • Tested code reading (checker.tcl proc image, ChangeLog 2010-06-08)
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment doctools has had the [image] markup since 1.4.7 (2010). Diagrams written with tklib's diagram package are rendered to images and embedded that way, as tcllib's pt docs do.
    ⬜1627868fff2007RFE term::vt100 control sequencesDONE (term::ansi::code::ctrl, added with the term module in c45a25f68f 2006-07-11, before this RFE)

    Open ticket 1627868fff · RFE · term · filed 2007-01-04

    Full verdict

    DONE (term::ansi::code::ctrl, added with the term module in c45a25f68f 2006-07-11, before this RFE)

    • Module term::ansi::code::ctrl 0.4 (term module)
    • Tested compared the attached VT100 list (vt100.txt) with ansi_cctrl.man. Cursor movement and positioning, save/restore cursor (with attributes), scroll region, erase line/screen, tab set/clear, device/cursor status queries, reset, line wrap, G0/G1 charset selection, graphics charset and SGR attributes are all there.
    • Cause n/a. Still missing are only rarely used modes: origin mode, cursor-key/keypad application mode, ESC D/M/E, double-height/width lines, printer control and reverse-screen. They can be built with term::ansi::code::define.
    • Prototype none (RFE)
    • Suggested action close as Fixed/Done
    • Suggested ticket comment The term module's term::ansi::code::ctrl package provides the VT100/ANSI control sequences: cursor movement, scroll regions, erase, tabs, charsets, queries and attributes. Anything still missing can be defined with term::ansi::code::define.
    ⬜1584520fff2006add option to tcllib's configureDONE (3ea92fe7fb, 2007-08-21: make test uses verbose batch mode when stdout is not a tty; make test_batch forces it)

    Open ticket 1584520fff · Patch · A Category Is Missing · filed 2006-10-25

    Full verdict

    DONE (3ea92fe7fb, 2007-08-21: make test uses verbose batch mode when stdout is not a tty; make test_batch forces it)

    • Module Makefile.in
    • Tested read Makefile.in test/test_batch/test_interactive targets
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed/Out of date
    • Suggested ticket comment Since 2007 make test uses the plain log when output is not a terminal, and make test_batch forces it. No configure option is needed.
    ⬜1492630fff2006Installer not creating tcllibc + related uuid issueDONE (make install runs install-binaries, i.e. sak.tcl critcl, which builds and installs tcllibc. devdoc/tcllib_installer.man ("Build & Insta…

    Open ticket 1492630fff · Bug · Installer · filed 2006-05-21

    Full verdict

    DONE (make install runs install-binaries, i.e. sak.tcl critcl, which builds and installs tcllibc. devdoc/tcllib_installer.man ("Build & Installation", parts/b_critcl.inc) documents building the accelerators.)

    • Module installer / Makefile.in / uuid 1.0.x
    • Tested code/doc read (Makefile.in install target, devdoc/tcllib_installer.man).
    • Cause n/a. uuid still uses generate_c only if critcl is already loaded, or through tcllibc, which is the intended design.
    • Prototype none
    • Suggested action close as Fixed/Out of date
    • Suggested ticket comment make install now builds and installs tcllibc via critcl (install-binaries), and the installer guide documents how to build the accelerators. uuid picks up the C version from tcllibc.
    ⬜1465066fff2006logger does not print correct args when passed multiple argsFIXED since tcllib 1.12 (cff1860c31, 2008-12-17, logger 0.9, RFE 2381524)

    Open ticket 1465066fff · Bug · logger · filed 2006-04-05

    Full verdict

    FIXED since tcllib 1.12 (cff1860c31, 2008-12-17, logger 0.9, RFE 2381524)

    • Module logger 0.9.5
    • Tested ${f}::debug foo bar baz printed [bar] [debug] 'baz' with tcllib 1.9, 1.10 and 1.11 (logger 0.7/0.8). With 1.12, 1.15, 1.19 and trunk on 8.6 and 9.1 it prints [foo] [debug] 'foo bar baz'.
    • Cause the old service-name tracking took arguments as the service. The 0.9 rewrite only treats the -_logger::service marker that way.
    • Prototype none
    • Suggested action close as Fixed.
    • Suggested ticket comment Works since logger 0.9 (tcllib 1.12, cff1860c31). Multiple arguments are now logged as one message.
    ⬜1438365fff2006possible error condition when port in useFIXED since tcllib 1.11 (7b5c1024fa, 2008-02-29, comm 4.5.7, fix for SF 1861565)

    Open ticket 1438365fff · Bug · comm · filed 2006-02-24

    Full verdict

    FIXED since tcllib 1.11 (7b5c1024fa, 2008-02-29, comm 4.5.7, fix for SF 1861565)

    • Module comm 4.7.3
    • Tested 9.1, 8.6: comm::comm new test -port <busy> -listen 1 and comm::comm config -port <busy> both give "couldn't open socket: address already in use"; test comm-5.0 covers it
    • Cause commConfigure renamed $chan unconditionally; now guarded by [info proc $chan]
    • Prototype none (fixed)
    • Suggested action close as Fixed (duplicate of 1861565)
    • Suggested ticket comment Fixed in comm 4.5.7 (tcllib 1.11) with the fix for 1861565fff: the error is now "address already in use"; test comm-5.0.
    ⬜1430694fff2006smtp or socket issue causes error 400 in Windows XP ProFIXED since tcllib 1.21 (ce8dab3154, 2021-03-22)

    Open ticket 1430694fff · Bug · smtp · filed 2006-02-13

    Full verdict

    FIXED since tcllib 1.21 (ce8dab3154, 2021-03-22)

    • Module mime / smtp 1.5.2
    • Tested sent 10 KB+ messages with -usetls 0 to a local fake SMTP server on 9.1/8.6 (Trf not available, so the non-Trf path). OK.
    • Cause The reports point at the Trf transform path with a non-blocking socket ("error writing sock: invalid argument", data cut off). Ticket 7849ff6729 fixed exactly this in ce8dab3154: the channel now runs blocking while the Trf transform is active, and no data is lost at the end. The rest is Windows XP / Tcl 8.4 era.
    • Prototype none
    • Suggested action close as Fixed (or Out of Date)
    • Suggested ticket comment The Trf path that cut off large messages was fixed in ce8dab3154 (ticket 7849ff6729, tcllib 1.21). Closing.
    ⬜1328920fff2005Symbolic differentiationDONE (8437472089, 2010-10-20, kennykb)

    Open ticket 1328920fff · RFE · math · filed 2005-10-17

    Full verdict

    DONE (8437472089, 2010-10-20, kennykb)

    • Module math, math::calculus::symdiff 1.0.2
    • Tested symdiff.tcl/.man/.test exist on trunk.
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Added as math::calculus::symdiff in 2010 (tcllib 1.13).
    ⬜1292966fff2005Performance: connection timeoutFIXED since tcllib 1.14 (52c11ba5ad, 2011-08-09, bug 3325112)

    Open ticket 1292966fff · Bug · ftp · filed 2005-09-16

    Full verdict

    FIXED since tcllib 1.14 (52c11ba5ad, 2011-08-09, bug 3325112)

    • Module ftp 2.4.14
    • Tested code read. The fconfigure -peername call in OpenControlConn was removed by that check-in, which is exactly the reporter's diagnosis.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment The slow -peername lookup in OpenControlConn was removed in ftp 2.4.11 (bug 3325112, tcllib 1.14).
    ⬜1144466fff2005RFE: logger: customizable output formatDONE (logger::utils / logger::appender, 81c17070d3, 2005-09-28, tcllib 1.8)

    Open ticket 1144466fff · RFE · log · filed 2005-02-19

    Full verdict

    DONE (logger::utils / logger::appender, 81c17070d3, 2005-09-28, tcllib 1.8)

    • Module logger 0.9.5, logger::utils 1.3.2, logger::appender 1.4
    • Tested logger::utils::applyAppender -appender console -serviceCmd $log -appenderArgs {-conversionPattern {\[%d\] \[%c\] \[%M\] \[%p\] %m}} prints [2026/11/15 01:06:52] [foo] [myproc] [warn] hello on 9.1 and 8.6. That includes the caller proc (%M), as the ticket asked. fileAppend -outputChannel covers the per-channel request in the comments.
    • Cause n/a. The %d date format is fixed rather than configurable, which is a small gap.
    • Prototype none
    • Suggested action close as Fixed/Out of date.
    • Suggested ticket comment Covered by logger::utils::applyAppender with -conversionPattern (%d %c %M %p %m ...) since tcllib 1.8. Output channels can be chosen with the fileAppend appender.
    ⬜1116123fff2005ncgi.tcl: misunderstands Content-Type: (bug #547274)FIXED since tcllib 1.9 (bd21455bc5, 2006-07-02, ncgi 1.3.1 "Fixed bug 547274": the nvlist content-type patterns became `application/x-www-form-urle…

    Open ticket 1116123fff · Patch · ncgi · filed 2005-02-04

    Full verdict

    FIXED since tcllib 1.9 (bd21455bc5, 2006-07-02, ncgi 1.3.1 "Fixed bug 547274": the nvlist content-type patterns became application/x-www-form-urlencoded* etc.)

    • Module ncgi 1.4.6
    • Tested ncgi::reset "a=b&c=d" "application/x-www-form-urlencoded; charset=UTF-8"; ncgi::nvlist gives a b c d on 9.1 and 8.6
    • Cause was the exact match on the type in nvlist
    • Prototype none. The attached diff is not applicable: it strips the parameters from [type], so multipart would lose its boundary.
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed in ncgi 1.3.1 (tcllib 1.9). Content-Type parameters such as "; charset=..." are now accepted for urlencoded and text/xml data.
    ⬜1104059fff2005Allow easy use of TLS with pop3DONE (-socketcmd tls::socket for pop3s in 1e4e56a154, 2009-04-13, pop3 1.7; STLS via -stls/-tls-callback in 38c311e999, 2011-01-23, pop3 1.8)

    Open ticket 1104059fff · RFE · pop3 · filed 2005-01-17

    Full verdict

    DONE (-socketcmd tls::socket for pop3s in 1e4e56a154, 2009-04-13, pop3 1.7; STLS via -stls/-tls-callback in 38c311e999, 2011-01-23, pop3 1.8)

    • Module pop3 1.11
    • Tested code and docs read; cannot test TLS here (no tcltls package).
    • Cause n/a. Follow-up worth a look: the STLS path calls tls::import ... -ssl2 0 -ssl3 1 -tls1 1 -cipher SSLv3,TLSv1, which is dated and may fail against modern TLS 1.2+-only servers with OpenSSL 3.
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Implemented: use -socketcmd tls::socket for pop3s (pop3 1.7) or -stls 1 for STARTTLS (pop3 1.8).
    ⬜1008619fff2004htmlparse::2tree gets the HTML Doc structure wrongFIXED since tcllib 1.9 (ab08ecbe1c, 2006-01-18)

    Open ticket 1008619fff · Bug · htmlparse · filed 2004-08-13

    Full verdict

    FIXED since tcllib 1.9 (ab08ecbe1c, 2006-01-18)

    • Module htmlparse 1.2.3
    • Tested the ticket's example on trunk (9.1, 8.6): <p> is now a sibling of <h1>. The htmlparse.tcl before ab08ecbe1c makes <p> a child of <h1>; the version from ab08ecbe1c gives the right tree.
    • Cause the post-pass Reorder was too aggressive. ab08ecbe1c ("less aggressive", SF Patch 953854) fixed it.
    • Prototype none (already fixed)
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed by ab08ecbe1c (SF patch 953854): <p> after <h1> is now its sibling.
    ⬜991959ffff2004command to send headersDONE (ncgi::header ?type? ?name value ...? already existed in 2004, before this RFE was filed (it is in the 2004-05-04 revision 35934da2af); it p…

    Open ticket 991959ffff · RFE · ncgi · filed 2004-07-15

    Full verdict

    DONE (ncgi::header ?type? ?name value ...? already existed in 2004, before this RFE was filed (it is in the 2004-05-04 revision 35934da2af); it prints Content-Type, extra headers and the pending Set-Cookie lines)

    • Module ncgi 1.4.6
    • Tested code reading; ncgi-12.x tests cover ncgi::header
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed / Works as designed
    • Suggested ticket comment ncgi::header (and ncgi::redirect, which uses it) does exactly this. A compatibility layer for Rivet/tclhttpd can redefine it.
    ⬜880692ffff2004ncgi - = in cgi value breaks parsing of variable/value pairsFIXED since tcllib 1.9 (6ad5a29080, 2006-07-02, ncgi 1.3.2; the greedy (.*)=(.*) regexp was replaced by string first =)

    Open ticket 880692ffff · Patch · ncgi · filed 2004-01-20

    Full verdict

    FIXED since tcllib 1.9 (6ad5a29080, 2006-07-02, ncgi 1.3.2; the greedy (.*)=(.*) regexp was replaced by string first =)

    • Module ncgi 1.4.6
    • Tested ncgi::reset "a=b=c&d==&=x&e=1" ""; ncgi::nvlist gives a b=c d = anonymous x e 1 on 9.1 and 8.6
    • Cause was ncgi.tcl:nvlist greedy regexp, which split at the last =
    • Prototype none (already fixed)
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed in ncgi 1.3.2 (tcllib 1.9): nvlist now splits at the first "=", so "a=b=c" gives name a, value "b=c".
    ⬜858218ffff2003errors thrown with no explanationFIXED since tcllib 1.6 (9bd4ab9e64, 2003-12-01)

    Open ticket 858218ffff · Bug · ftp · filed 2003-12-11

    Full verdict

    FIXED since tcllib 1.6 (9bd4ab9e64, 2003-12-01)

    • Module ftp 2.4.14
    • Tested code read. The regexp is now ^-?([0-9]+)( |-)?(.*)$ with the missing negation fixed, and the message (which includes the line) is only shown in VERBOSE mode. Normal servers (ftpd) parse fine.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Fixed in December 2003 (ftp ChangeLog: wrong condition and an extra caret in the regexp). This was released in tcllib 1.6.
    ⬜772489ffff2003doctools: latex, add document internal cross-referencingDONE (doctools 1.3.4, 2008-05-15: (sub)section ids plus [sectref id]. fmt.latex emits \label{id} after each \section and \ref{label} for se…

    Open ticket 772489ffff · RFE · doctools · filed 2003-07-16

    Full verdict

    DONE (doctools 1.3.4, 2008-05-15: (sub)section ids plus [sectref id]. fmt.latex emits \label{id} after each \section and \ref{label} for sectref.)

    • Module doctools 1.6.1 fmt.latex
    • Tested code reading, fmt.latex lines 147-148 and 337-339
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Since doctools 1.3.4 sections and subsections can have ids, and [sectref] refers to them. The LaTeX engine renders these as \label and \ref.
    ⬜747674ffff2003ftp_lib.tclFIXED since tcllib 1.10 (cc984c045f, 2006-11-09, bug 1191607)

    Open ticket 747674ffff · RFE · ftp · filed 2003-06-02

    Full verdict

    FIXED since tcllib 1.10 (cc984c045f, 2006-11-09, bug 1191607)

    • Module ftp 2.4.14
    • Tested ftp::Quote $h HELP against ftpd returns the whole 214 multi-line reply (9.1).
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Since ftp 2.4.7 (bug 1191607), ftp::Quote returns the complete multi-line reply.
    ⬜732597ffff2003move tcllib doc generation to build stepDONE (3286b5637f 2013-02-15 "installer uses the embedded documentation instead of generating", then 888d8ab7b9/46a48af79f which moved the ins…

    Open ticket 732597ffff · RFE · general request · filed 2003-05-05

    Full verdict

    DONE (3286b5637f 2013-02-15 "installer uses the embedded documentation instead of generating", then 888d8ab7b9/46a48af79f which moved the installed docs to idoc/; first release tcllib 1.16)

    • Module installer / sak
    • Tested installer.tcl now copies pregenerated nroff/html from idoc/; nothing is generated at install time.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Done: since tcllib 1.16 the documentation is pregenerated in idoc/ and the installer only copies it.
    ⬜704219ffff2003doctools - HTML customization featuresDONE (mostly). Since doctools 1.3.x (2008-05, ChangeLog "_html.tcl ... divisions a CSS can lock onto") the HTML output puts every part in a div/`…

    Open ticket 704219ffff · RFE · doctools · filed 2003-03-15

    Full verdict

    DONE (mostly). Since doctools 1.3.x (2008-05, ChangeLog "_html.tcl ... divisions a CSS can lock onto") the HTML output puts every part in a div/span with a doctools_* class. The engine parameters meta, header, footer, xref and raw let you add your own CSS. Section headings are now "Name"/"Description" in initial caps. You still cannot drop the Name/Description sections or change the tags.

    • Module doctools 1.6.1 fmt.html
    • Tested generated HTML on 9.1, checked for class="doctools_section" etc. Read fmt.html __var.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment The HTML engine now emits classed divs/spans (doctools_section, doctools_synopsis, ...), and a stylesheet can be injected through the meta parameter, so fonts, colors and heading styles are customizable with CSS. Further structural changes need a custom engine.
    ⬜672957ffff2003New submodule: optimisationDONE (e69e86b94f, 2003-04-18, Arjen's optimisation procs; solveLinearProgram followed)

    Open ticket 672957ffff · RFE · math · filed 2003-01-23

    Full verdict

    DONE (e69e86b94f, 2003-04-18, Arjen's optimisation procs; solveLinearProgram followed)

    • Module math, math::optimize 1.0.2
    • Tested the package exists with docs and tests.
    • Prototype none
    • Suggested action close as Fixed
    • Suggested ticket comment Included as math::optimize in 2003, including solveLinearProgram.
    ⬜618241ffff2002ftp::Get <source> -variable <var> bombsFIXED since tcllib 1.4 (0515c88443, 2002-08-06, SF bug 582668)

    Open ticket 618241ffff · Bug · ftp · filed 2002-10-03

    Full verdict

    FIXED since tcllib 1.4 (0515c88443, 2002-08-06, SF bug 582668)

    • Module ftp 2.4.14
    • Tested ftpd on 127.0.0.1. ftp::Get $h a.txt -variable v returns 1 with correct content in active and passive mode, on 9.1 and 8.6. The $dest check is now if {$ftp(inline) == 0}, exactly the fix suggested in the ticket.
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed (duplicate of 582668)
    • Suggested ticket comment This was fixed as SF bug 582668 in 2002 (ftp 2.4, tcllib 1.4); Get -variable works on current trunk.
    ⬜516723ffff2002doctools - javadoc like applicationDONE (mkdoc module + app added in tcllib 1.21, 8defde4552, 2022-04-28; also tepam::doc_gen for TEPAM procs)

    Open ticket 516723ffff · RFE · doctools · filed 2002-02-12

    Full verdict

    DONE (mkdoc module + app added in tcllib 1.21, 8defde4552, 2022-04-28; also tepam::doc_gen for TEPAM procs)

    • Module none (RFE for new app)
    • Tested checked that modules/mkdoc and apps/mkdoc exist on trunk
    • Cause n/a
    • Prototype none (RFE)
    • Suggested action close as Fixed
    • Suggested ticket comment Tcllib 1.21 added the mkdoc package and application. They extract documentation comments from Tcl source files (Markdown, not doctools). For TEPAM procedures there is also tepam::doc_gen.
    ⬜486547ffff2001note for future struct::graph developmenDONE (a C implementation of struct::graph, the main topic of the discussion, is in tcllib: modules/struct/graph/*.c, graph_c.tcl. Graph I/O formats…

    Open ticket 486547ffff · RFE · struct :: graph · filed 2001-11-28

    Full verdict

    DONE (a C implementation of struct::graph, the main topic of the discussion, is in tcllib: modules/struct/graph/*.c, graph_c.tcl. Graph I/O formats are not part of struct.)

    • Module struct::graph 2.4.4
    • Tested code inspection
    • Cause n/a
    • Prototype none
    • Suggested action close as Fixed / Out of Date
    • Suggested ticket comment struct::graph now has a critcl C implementation, and the external cgraph link is dead. The CPAN pointer is just a note. Closing.
    ⬜473656ffff2001pop3::retrieve blocking foreverFIXED since ~tcllib 1.1/1.6 (the -retr-mode/size handling of 2001-07-31 and the binary/dot-unstuffing rework of 2003-11-18, see pop3 ChangeLog). Th…

    Open ticket 473656ffff · Bug · pop3 · filed 2001-10-22

    Full verdict

    FIXED since ~tcllib 1.1/1.6 (the -retr-mode/size handling of 2001-07-31 and the binary/dot-unstuffing rework of 2003-11-18, see pop3 ChangeLog). The report was against pop3 1.2.

    • Module pop3 1.11
    • Tested 9.1, 8.6, tclsh and wish (Xvfb :117): retrieve against a local fake POP3 server. Dot-stuffed lines are unstuffed, and a short read (more data than announced) completes. A server announcing *more* octets than it sends still blocks in read (RetrFast); -retr-mode slow is the workaround. The pop3 test suite passes (44/45, the 1 skip is TLS).
    • Prototype none
    • Suggested action close as Fixed (out of date)
    • Suggested ticket comment Both problems (blocking read of the announced size, and missing dot-unstuffing) were fixed in later pop3 versions (2001/2003); this report was about pop3 1.2.