Untouched open Tk bug tickets: Docs

Triage of 548 open Bug tickets: those with no comment or check-in from me on 2026-09-24, plus 38 added on 2026-10-04 (filed since then, Pending then, or reopened; 16 of them filed by me), plus 68 added on 2026-10-22 (all other open or Pending Bug tickets not filed by me: the 30 created before 8.5, 31 I had already worked on, and 7 filed or missed since 10-04). Reproduced on Linux/X11 under Xvfb (main 9.1 tip, 9.0, 8.7, 8.6 tip, 8.5, 8.4) and on a Windows 11 VM (9.1, 8.6, 8.5, 8.4). Rows marked patch link to a prototype patch (150 in all, in tk-triage/prototypes). Updated 2026-11-11 with what happened since: the emoji before each ticket id shows its progress. On 2026-10-24 every prototype for a still-open ticket was verified again on current main, on Linux and Windows, and given a test where practical; on 2026-10-25 seven more reproduced tickets got prototypes.

Docs 11

⬜1a15f7cc4d2025Unclear docs on MacTypes in tk_getOpenFile (patch included)patchDOCS (still unchanged in main doc/getOpenFile.n)

Open ticket 1a15f7cc4d · 37. [tk_get*File] · filed 2025-04-22

Full verdict

DOCS (still unchanged in main doc/getOpenFile.n)

  • Platform docs
  • Tested source check. macosx/tkMacOSXDialog.c uses only clausePtr->patterns (extensions) and never reads macTypes, so macTypes are ignored on every platform.
  • Cause the doc still describes classic Mac OS AND/OR matching by macType.
  • Prototype 1a15f7cc4d.patch (30 lines, docs only). It says macType is accepted but ignored on all platforms, drops the Macintosh AND/OR paragraphs, and removes TEXT/GIFF from the example.
  • Suggested ticket comment Current macOS Tk ignores macTypes: the dialog code filters by extension only. A doc update is prepared saying so and cleaning up the example.
✅2903b4c9ab2025Undocumented constraints on menubarsDOCS; the macOS breakage is CANNOT-TEST (macOS). NOT-REPRODUCED on X11.

Open ticket 2903b4c9ab · 10. Generic Menus · filed 2025-03-21

✅ Since triage (updated 2026-11-11)

Status now: Closed (Fixed)

  • 🔧 2026-10-11 check-in 7d68f657d9 on menubar-clone-recreate, in main, 9.0: Fix 2903b4c9ab: a toplevel lost its menubar when the menubar clone was destroyed, e.g. together with the other children of the toplevel, and setting -menu to
  • 💬 2026-10-11 serhiy.storchaka: The menubar breaks because the script destroys all children of the toplevel, including the menubar clone .tl.#menuBar, and setting -menu to the same value did not create the clone again. This happens on X11 too, and the menubar does not need to be a child of the toplevel. Proposed fix in branch menubar-clone-recreate, test menu-25.17.
  • 💬 2026-10-31 nemethi: Merged by Jan into main and core-9-0-branch by commits 5b1e8d51 and [b5da3af6 ]. Backported to core-8-6-branch by commit 389e0875. Closing the ticket.
Full verdict

DOCS; the macOS breakage is CANNOT-TEST (macOS). NOT-REPRODUCED on X11.

  • Platform docs (the failure itself is macOS)
  • Tested X11 main and 8.6. A menubar .mb (child of .) attached to two toplevels .t and .u works: clones .t.#mb/.u.#mb are mapped and cascades post. doc/menu.n already documents the cascade-must-be-child rule (lines 189, 685). The "menubar must be a child of its toplevel" rule is not documented; the generic/tkMenu.c header comment says clones handle it.
  • Cause on X11 (Windows presumably the same, untested) the menubar clone mechanism makes the constraint unnecessary. The macOS failure needs investigation there (it is probably in how macosx/tkMacOSXMenu.c finds the menubar clone).
  • Prototype none (whether to document a macOS-only restriction or fix Aqua needs a Mac)
  • Suggested ticket comment On X11 a menubar that is not a child of its toplevel works, even when shared between toplevels, thanks to clones. So the restriction is Aqua-specific: either fix the Aqua menubar code or add a note to the MENUBARS section of menu.n.
⬜2e07480f5f2020ttk::spinbox needs to document <MouseWheel> bindingpatchDOCS (still missing on main: ttk_spinbox.n has no DEFAULT BINDINGS section)

Open ticket 2e07480f5f · 88. Themed Tk · filed 2020-05-28

Full verdict

DOCS (still missing on main: ttk_spinbox.n has no DEFAULT BINDINGS section)

  • Platform docs
  • Tested main (read library/ttk/spinbox.tcl; checked bind .sb <MouseWheel> break disables wheel spinning on X11, main)
  • Cause doc/ttk_spinbox.n only documents <<Increment>>/<<Decrement>>. Differences from TEntry: arrow press/repeat, Double-1 selects all, Triple-1 disabled, Up/Down, MouseWheel/Option-MouseWheel/TouchpadScroll spin the value, Shift-MouseWheel ignored.
  • Prototype 2e07480f5f.patch (45 lines, doc only: new "DEFAULT BINDINGS" section listing the differences from ttk::entry plus how to disable the wheel)
  • Suggested ticket comment Added a DEFAULT BINDINGS section to ttk_spinbox.n that lists the differences from ttk::entry (arrows, double/triple click, Up/Down, mouse wheel) and shows how to disable wheel spinning with bind .sb <MouseWheel> break.
⬜835e5e17ec2019missing manual page for ttk::cursorDOCS (still undocumented on main. ttk::cursor is defined in library/ttk/cursors.tcl and is used internally by several ttk widgets)

Open ticket 835e5e17ec · 88. Themed Tk · filed 2019-12-01

Full verdict

DOCS (still undocumented on main. ttk::cursor is defined in library/ttk/cursors.tcl and is used internally by several ttk widgets)

  • Platform docs
  • Tested main (grep doc/)
  • Cause n/a. This is effectively an RFE to make ttk::cursor a public API. See the analysis in b3b56ae8dc.
  • Prototype none (making the command public is an API decision; a man page would list the symbolic names from the cursors.tcl table and the per-platform mapping)
  • Suggested ticket comment ttk::cursor is still internal and undocumented. Documenting it would make it public API (it needs a decision, maybe a TIP), so this is an RFE rather than a bug.
✅e02aac5cab2017Grab documentation does not have caveats for windows & macpatchDOCS. The proposed paragraph was partly wrong and has been revised.

Open ticket e02aac5cab · 56. [grab] · filed 2017-02-20

✅ Since triage (updated 2026-11-11)

Status now: Closed (Fixed)

  • 🔧 2026-10-08 check-in e2d44be67a on grab-doc-windows, in main, 9.0: Fix e02aac5cab: document the behaviour of grabs on Windows: a global grab does not lock out other applications, and a toplevel window outside the grab cannot
  • 🔧 2026-11-03 check-in 6d5d5bcd05 on trunk, in main, 9.0: Fix for e02aac5cab: Grab documentation does not have caveats for windows & mac
  • 🔧 2026-11-03 check-in e7befe1430 on core-9-0-branch, in main, 9.0, 8.6: Fix for e02aac5cab: Grab documentation does not have caveats for windows & mac
  • 🔧 2026-11-03 check-in d9a6bca1b2 on core-8-6-branch, in 8.6: Fix for e02aac5cab: Grab documentation does not have caveats for windows & mac
  • 🔧 2026-10-08 check-in 01ff9af74e on grab-doc-windows, in main, 9.0: Correct the Windows grab paragraph: an outside toplevel can be raised but not activated, and cannot be moved during a global grab (checked on Windows 11).
  • 💬 2026-10-08 serhiy.storchaka: The macOS behaviour has been documented since 1edd8b7a. Proposed documentation of the Windows behaviour in branch grab-doc-windows: a global grab does not lock out other applications, and a toplevel outside the grab cannot be activated or raised, its activation gives the focus to the grab window.
  • 💬 2026-10-08 serhiy.storchaka: Checked on Windows 11 and corrected the paragraph in grab-doc-windows: a toplevel outside the grab can be raised, but not activated, and during a global grab it cannot be moved either.
  • 💬 2026-11-03 nemethi: Serhiy, many thanks! Merged into main, core-9-0-branch, and core-8-6-branch by commits 6d5d5bcd, [e7befe1], and d9a6bca1.
Full verdict

DOCS. The proposed paragraph was partly wrong and has been revised.

  • Platform docs (Windows)
  • Tested read doc/grab.n on main, win/tkWinPointer.c (XGrabPointer/XGrabKeyboard are simulated inside the process, TkpSetCapture uses SetCapture) and win/tkWinWm.c WmProc (WM_ACTIVATE, WM_NCHITTEST, WM_SYSCOMMAND grab handling).
  • Cause missing documentation
  • Prototype e02aac5cab.patch (doc/grab.n, +7 lines). New Windows paragraph after the macOS one: a global grab does not lock out other applications; during any grab, toplevels outside the grab tree cannot be activated, and window manager commands on them other than move, size and restore are ignored; toplevels in the grab tree other than "." cannot be minimized. The wording comes from reading the code, so check it on Windows with win/e02aac5cab.tcl (interactive, prints a VERDICT line).
  • Suggested ticket comment The macOS caveats are documented since 1edd8b7a16. A Windows paragraph for grab.n is prototyped (a global grab does not lock out other applications; restrictions on toplevels outside the grab tree) and needs checking on Windows.
Windows VM, run 3
  • Platform docs (Windows)
  • Verdict DOCS. The proposed paragraph was partly wrong and has been revised.
  • Tested work-win/e02_auto.tcl on ref91 with real clicks, drags and posted WM_SYSCOMMANDs, plus a control run without a grab (where every action works, except the corner resize drag, which my automation could not do even without a grab, so resizing is not verified):
    • Global grab: another process still receives clicks and keys (confirmed). Outside toplevel: a click does not activate it (the grab window stays foreground); a title-bar drag does NOT move it; the minimize/maximize/close buttons are ignored; SC_MINIMIZE/SC_MAXIMIZE/SC_CLOSE are ignored; SC_RESTORE of an iconified outside toplevel is ALSO ignored. A grab-tree toplevel other than .: minimize is ignored, and SC_MAXIMIZE is ignored too. . outside the grab tree: minimize is ignored.
    • Local grab: the same, except that a title-bar drag moves the outside window, SC_RESTORE restores it, and SC_MAXIMIZE on the grab window works.
    • Grab on .: . can be minimized.
  • Cause missing docs. Behaviour comes from tkWinWm.c WM_SYSCOMMAND/WM_NCHITTEST/WM_MOUSEACTIVATE; with a global grab, the capture on the grab window also blocks mouse moves of other windows.
  • Prototype e02aac5cab.patch (doc/grab.n, +9 lines, revised): other apps are not locked out by a global grab; outside toplevels cannot be activated and their minimize/maximize/close are ignored; during a local grab they can still be moved, resized and restored, but during a global grab they cannot be moved or restored either; grab-tree toplevels other than . cannot be minimized.
  • Suggested ticket comment Windows behaviour was checked on Windows 11. A grab.n paragraph documenting it is prepared: other applications are not blocked, windows outside the grab cannot be activated, minimized, maximized or closed, and during a global grab they cannot even be moved or restored.
⬜27961a0f8b20158.5 doc missing info on ".window" menu on OS XDOCS. Still missing on core-8-5-branch; 8.6 and main document it.

Open ticket 27961a0f8b · 11. Aqua Menus · filed 2015-06-22

Full verdict

DOCS. Still missing on core-8-5-branch; 8.6 and main document it.

  • Platform docs
  • Tested doc/menu.n on 8.5, 8.6, main. core-8-5-branch macosx/tkMacOSXMenu.c does support .window (ENTRY_WINDOWS_MENU).
  • Cause the 8.6 menu.n paragraph on .menubar.window was never backported to 8.5.
  • Prototype none. The fix would be to copy the ".menubar.window" sentence and paragraph from the 8.6 menu.n into the 8.5 one, but 8.5 is EOL.
  • Suggested ticket comment Documented in 8.6 and later. 8.5 is end-of-life and will not get further doc releases, so closing as Out of Date.
⬜9daafa27a92015Doc improvement for Tk "image photo" supportpatchDOCS (the premise is wrong for core Tk)

Open ticket 9daafa27a9 · 41. Photo Images · filed 2015-04-23

Full verdict

DOCS (the premise is wrong for core Tk)

  • Platform docs
  • Tested main, 8.6 tip, 8.6.8, 8.6.9, 8.6.13, 8.5. Built-in data -format png|gif|ppm returns binary data, not base64, in all of them (8.5 has no string writer for png/gif). put/-data take binary for all three, and also base64 for PNG and GIF only. The double encoding the reporter saw comes from extension handlers (Img), not from the core.
  • Cause doc/photo.n says what -data accepts (binary, or base64 for some formats), but it doesn't say what data returns or what put accepts.
  • Prototype 9daafa27a9.patch (6 changed lines in doc/photo.n). It says the built-in handlers return binary data, points to binary encode base64, and says put accepts binary data or base64 (PNG/GIF), as -data does.
  • Suggested ticket comment In current Tk (8.6, 9.x) the built-in PNG/GIF/PPM handlers return binary data from ''data -format'', and ''put''/''-data'' accept either binary or (PNG/GIF) base64 data. So there is no double encoding in the core; Img handlers may behave differently. photo.n could state this more explicitly (small doc patch prepared).
⬜2930718fff2010Undocumented API: Tk_OutlinepatchDOCS (still undocumented on main: no doc/*.3 mentions Tk_CreateOutline etc.)

Open ticket 2930718fff · 04. Canvas Basics · filed 2010-01-12

Full verdict

DOCS (still undocumented on main: no doc/*.3 mentions Tk_CreateOutline etc.)

  • Platform docs
  • Tested grep of doc/ on main and 8.6
  • Cause public stubs 221-226 (Tk_CreateOutline, Tk_DeleteOutline, Tk_ConfigOutlineGC, Tk_ChangeOutlineGC, Tk_ResetOutlineGC, Tk_CanvasPsOutline) never got a man page.
  • Prototype 2930718fff.patch (new doc/CanvOutline.3, ~100 lines, written from tkCanvUtil.c and tkRectOval.c usage; plus Tk_CreateOutline added to SEE ALSO of CrtItemType.3; renders cleanly with groff -ww)
  • Suggested ticket comment Draft man page CanvOutline.3 documenting the six Tk_Outline functions and their use in item create/configure/display/delete/postscript procedures.
💬940220ffff2004wm geometry contradictions (and inadequately documentedpatchDOCS. Behaviour is now consistent. On X11 (fvwm, openbox) wm geometry . 200x100+50+60 gives a 200x100 interior, excluding the menubar and decorat…

Open ticket 940220ffff · 62. Geometry Functions · filed 2004-04-22

💬 Since triage (updated 2026-11-11)

Status now: Open

  • 💬 2026-09-30 serhiy.storchaka: See [220791] and d1a711b500. The proposed fix for d1a711b500 makes wm geometry report the position of the decorated window also under KWin on Wayland (Xwayland), which decorates windows without reparenting them. The offsets under metacity and mutter after wm withdraw and wm deiconify, caused by invisible borders of their frames, are not addressed.
Full verdict

DOCS. Behaviour is now consistent. On X11 (fvwm, openbox) wm geometry . 200x100+50+60 gives a 200x100 interior, excluding the menubar and decorations, with the decorated frame at 50,60. This matches the Windows semantics described in the ticket. Without a WM only the menubar offset applies.

  • Platform docs
  • Tested main, 8.6 tip × no WM, fvwm, openbox (twm failed to start). Menubar configured in all runs.
  • Cause wm.n never says that width/height exclude the WM frame and the menubar, or what "window's border" means for x/y. A paragraph comparing it with winfo geometry has been added since, but it stays vague.
  • Prototype 940220ffff.patch (doc only, wm.n, adds 10 lines)
  • Suggested ticket comment The behaviour is now the same on X11 (for common WMs) and Windows: the size is the interior without decorations or the menubar, and the position is the outer frame. A doc clarification for wm.n is prepared.
⬜220807ffff2000Docs missing for Tk_MainEx APIpatchDOCS (still missing in main, 9.0, 8.7, 8.6, 8.5, 8.4: no man page mentions Tk_MainEx; Tk_Main is only a macro around Tk_MainEx since 8.4, and Tk_Ma…

Open ticket 220807ffff · 74. Application Embedding · filed 2000-11-01

Full verdict

DOCS (still missing in main, 9.0, 8.7, 8.6, 8.5, 8.4: no man page mentions Tk_MainEx; Tk_Main is only a macro around Tk_MainEx since 8.4, and Tk_MainExW exists on Windows since 8.7)

  • Platform docs
  • Tested grep of doc/ in all source trees; nroff rendering of the patched page
  • Cause doc/Tk_Main.3 documents only Tk_Main
  • Prototype 220807ffff.patch (67-line diff, ~28 changed lines; doc only): adds Tk_MainEx/Tk_MainExW to NAME/SYNOPSIS/ARGUMENTS (charargv, wideargv, interp) modelled on Tcl_Main.3, explains that Tk_Main is a macro calling Tcl_FindExecutable + Tcl_CreateInterp + Tk_MainEx, that Tk_MainExW is Windows-only, adds Tcl_Main(3) to SEE ALSO; also fixes argv type (char / wchar_t ) and -DUNICODE markup. For 8.6, drop Tk_MainExW and use int argc.
  • Suggested ticket comment Tk_MainEx (and Tk_MainExW on Windows) is still undocumented. A doc patch adding them to Tk_Main.3, modelled on Tcl_Main.3, is prepared.
⬜220812ffff2000Docs missing for Tk_UpdatePointer APIpatchDOCS (main, 9.0, 8.6)

Open ticket 220812ffff · 65. Generic Window Operations · filed 2000-11-01

Full verdict

DOCS (main, 9.0, 8.6)

  • Platform generic (implementation used on Windows/Aqua; no-op on X11)
  • Tested read the trunk code. Tk_UpdatePointer is public stub 184 in generic/tk.decls on all branches, and no doc/*.3 mentions it. Implementations: generic/tkPointer.c (pointer emulation: Enter/Leave via TkInOutEvents, ButtonPress/Release, implicit grab, TkpSetCapture) and unix/tkUnix.c (empty).
  • Cause doc/ never got a page for this public stub.
  • Prototype 220812ffff.patch (58 lines, new doc/UpdatePointer.3, renders cleanly with groff -man). It describes the arguments, the events generated, the grab handling, and the no-op on X11.
  • Suggested ticket comment Tk_UpdatePointer is still public (stub 184) and in use by the Windows/macOS ports, so it should be documented rather than hidden. A short man page is drafted.