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 Outdated, superseded, duplicate and misfiled tickets are listed separately.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.
.., cl or str. pt::pe verify still rejects [pt::pe class abc] and `[…DOCS (all points still valid). The PE serialization docs do not list .., cl or str. pt::pe verify still rejects [pt::pe class abc] and [pt::pe str abc] ("invalid operator"), while .. verifies. t x with multi-character x is not documented as invalid. ddigit is still described as regexp-based.
pt::pe verify on range/class/str results on 9.1. Grep of the docs; pexpression.inc has been unchanged since 2010.
pt::pe class/str and pt::pe verify
.., cl and str are undocumented, and verify rejects cl/str although pt::pe class/str produce them.
DOCS (the sentence is still in generator.man on trunk)
{{1} {2} {3}} and {1 2 3} are the same Tcl value; mkMatrix 3 1 returns 1.0 1.0 1.0, and shape gives 3)DOCS (inherent: {{1} {2} {3}} and {1 2 3} are the same Tcl value; mkMatrix 3 1 returns 1.0 1.0 1.0, and shape gives 3)
shape {{1} {2} {3}} gives 3, and shape [mkMatrix 3 1 1.0] gives 3.
DOCS (the reported getfiles problem is misuse, not a bug)
getfiles [list /path/a.zip] 0 returns the file. getfiles {/path/*.zip} 0 on Unix warns "no files match", as documented (no globbing on Unix). The reporter passed the pattern instead of the remaining argv. There is no optind in cmdline: argv is modified in place, and the example now says so (2dce7b748a, Fixed).
set files [cmdline::getfiles $argv 0] after getoptions.)
::pki::public_key keylist ?password? ?encodePem? (pki.tcl:861) and ::pki::pkcs::parse_public_key key ?password? (pki.tcl:1135) …DOCS (RFE-OPEN; ::pki::public_key keylist ?password? ?encodePem? (pki.tcl:861) and ::pki::pkcs::parse_public_key key ?password? (pki.tcl:1135) exist but pki.man still documents neither)
aycock::parser, which does not exist (`invalid command name "aycock::parser"…DOCS (still valid). The example is also wrong: it and the [call] use aycock::parser, which does not exist (invalid command name "aycock::parser"). The real command is grammar::aycock::parser.
::grammar::aycock::parser and replaces the example with a slightly condensed version of the submitted lexer plus the parser. I extracted the example from the .man and ran it: it prints 40 on 9.1 and 8.6, and dtplite renders it.
aycock::parser. Proposed doc update with a lexer and the correct grammar::aycock::parser name.
DOCS (still only oodialect.md, no .man; not in the generated docs. The oo::meta twin a20b080092 was fixed in 2019.)
DOCS (partly addressed: since tcllib 1.21 (900fb4ff6d, 2021-10-31) parse_cert/parse_csr say "can be either PEM or DER encoded". parse_key, verify_cert etc. still just say "key"/"cert", and nothing states that the arguments are data, not file names)
DOCS (fixed on trunk 2b2bcd7ecf, 2026-01-27). json.man documents dict-or-list results and the type ambiguity. DUPLICATE of 2df194042b in substance.
DOCS (fixed on trunk, 3ec5b08cd9 / b96bdd24d5, merged 2b2bcd7ecf, 2026-01-23/27, not yet released)
DOCS (RFE-OPEN; vt_otp.man still has no example)
rb) for keychanw/keychanr, call tcl::transform::otp $out $kw $kr, then fcopy. Reading back through the transform with fresh key channels gives the plaintext.
tcl::transform::otp $chan $keyw $keyr, then write or fcopy through $chan.
DOCS / RFE-OPEN (still true: mime::word_encode, word_decode, field_decode are not in mime.man; non-ASCII header values are written raw, not RFC 2047 encoded)
-header {Subject "Привіт öö"} → raw UTF-8 bytes in the header
DOCS
DOCS (works as designed; the manual is misleading)
set ftp::VERBOSE 1; set ftp::DEBUG 1 prints nothing by default. After log::lvSuppress debug 0 everything shows, on 9.1 and 8.6.
::ftp::DisplayMsg logs through log::log debug, and the log package suppresses debug/info/notice/warning by default. The suggested log::lvSuppressLE emergency 0 inside the ftp package would change the logging of the whole application, so it should not be done.
debug (errors at level error). The log package suppresses level debug by default, so an application setting VERBOSE or DEBUG also has to call log::lvSuppress debug 0, or use the -output option of ::ftp::Open."
log::lvSuppress debug 0 (or use -output). Changing log's global settings from inside ftp would affect the whole application, so we will document this instead.
option statement already states the rule; snitfaq.man does not)DOCS (snit.man's option statement already states the rule; snitfaq.man does not)
option transformBefore and option -transformBefore both give badly named option on 9.1 and 8.6. The rule (Comp.OptionNameIsValid: leading hyphen, no upper case, no space) is already covered by tests option-*.
DOCS (still present: two uses of "session" in the reset and urlStub descriptions)
DOCS: still no man page and no test suite. Duplicate 1954483fff was closed in favour of this ticket.
DOCS: still no man page. spf.test exists now; tcllib_dns.man only links to an SPF website.
[example {...}] doctools copies…DOCS (REPRODUCED on 9.1 and 8.6. The doubled backslashes are in the des.man source, not added by doctools. Inside [example {...}] doctools copies backslashes verbatim, so \\0 in the source shows up as \\0 in HTML/markdown/text. That is also wrong Tcl: string repeat \\0 8 is 16 literal characters, not 8 NUL bytes. aes.man has the same error, string repeat \\0 16, and its printed hex result only matches the single-backslash form. I checked: with \\0 the result is fdb1..., with \0 it is 66e94bd4... as documented.)
\\ inside a braced example. The only doctools special case is a backslash at the end of a line, where \\ gives one \ (AK's continuation-line note). That does not apply here.
DOCS (partly done: an "Uploading a file" EXAMPLES section was added in 2007, c33818a4c4, but there is still no basic read-a-form example)
sak.tcl doc validate ncgi passes.
DOCS
NList handle ?directory?, while List documents ?pattern?
pattern and notes that directory prefixes in the result depend on the server (see 1584235).
DOCS (the -part/-parts naming was fixed via 3354014/3552226; there is still no example of building a multipart message, or of naming an attachment with Content-Disposition/filename)