Untouched open Tk bug tickets: Not tested

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.

Needs Windows 21

⬜e3e32831782025MS-Win: tk_getOpenFile returns short path, if path is longer than 256 charactersNEEDS-WINDOWS (inconclusive; LongPathsEnabled=0 on the VM)

Open ticket e3e3283178 · 37. [tk_get*File] · filed 2025-12-05

Full verdict

NEEDS-WINDOWS (inconclusive; LongPathsEnabled=0 on the VM)

  • Platform Windows
  • Tested nothing on X11 (Win32 dialog only).
  • Cause Windows returns 8.3 names from the file dialog unless the process is long-path aware and long paths are enabled in the system.
  • Prototype none (already fixed in main through the manifest)
  • Repro win/e3e3283178.tcl. It creates a path over MAX_PATH under %TEMP%, opens tk_getOpenFile on it, presses Enter through WScript SendKeys, prints LongPathsEnabled, and prints PASS/FAIL.
  • Suggested ticket comment Fixed in 9.1 by the longPathAware manifest (0a9902c812), when long paths are enabled in Windows (LongPathsEnabled=1). Otherwise Windows itself returns the short name. Should the manifest change be backported to 9.0 and 8.6?
Windows VM, run 6
  • Platform Windows
  • Verdict NEEDS-WINDOWS (inconclusive; LongPathsEnabled=0 on the VM)
  • Tested A 313-character path was created, but the file dialog was not dismissed by the WScript SendKeys helper, so no path was returned. Long paths are disabled on the VM, so a short path is expected anyway.
  • Prototype none
  • Suggested ticket comment Needs a manual check with LongPathsEnabled=1.
⬜6854ee6a2e2023Test imgPhoto-7.3 causes X server to crash (vcxsrv)NEEDS-WINDOWS (VcXsrv). NOT-REPRODUCED on Xvfb.

Open ticket 6854ee6a2e · 39. Image Basics · filed 2023-04-21

Full verdict

NEEDS-WINDOWS (VcXsrv). NOT-REPRODUCED on Xvfb.

  • Platform X11 (VcXsrv on Windows, rootless mode)
  • Tested imgPhoto-7.* in the wt-a3 build (main): all 3 pass on Xvfb. The standalone repro with -sync gives a 32-bit frame visual inside a 24-bit root and finishes OK on main and 8.6.18.
  • Cause the server crashes in fbCopyWindowProc/memcpy (chrstphrchvz). Tk only creates a depth-32 child window inside a depth-24 parent, which is valid X11, so this looks like a VcXsrv rootless-mode bug with mixed-depth windows.
  • Prototype none (server bug)
  • Repro win/6854ee6a2e.tcl (run an X11 wish against VcXsrv in "Multiple windows" mode; if "RESULT: OK" never prints, the bug is reproduced)
  • Suggested ticket comment Does not reproduce on Xvfb. The crash is inside VcXsrv (fbCopyWindowProc) on a legal depth-32-in-depth-24 window hierarchy, so it is a VcXsrv bug. Suggest closing as third-party.
⏳ee397e0a552023clipboard tests fail on WindowspatchNEEDS-WINDOWS (code read on trunk, core-8-6-branch and the bug-ee397e0a55 branch; the VM was powered off, so nothing was run)

Open ticket ee397e0a55 · 52. [clipboard] · filed 2023-04-19

⏳ Since triage (updated 2026-11-11)

Status now: Pending (Fixed)

    Full verdict

    NEEDS-WINDOWS (code read on trunk, core-8-6-branch and the bug-ee397e0a55 branch; the VM was powered off, so nothing was run)

    • Platform Windows
    • Tested:
      • The ticket says Fixed/Pending, but the branch bug-ee397e0a55 (ed6b64623c, 4f8ad61cef, ffb5d8f227, 3a96613ed1) was never merged. Trunk still calls plain OpenClipboard in TkSelGetSelection, in UpdateClipboard (return value ignored) and in WM_RENDERALLFORMATS.
      • The branch only retried with Tcl_Sleep. Ashok saw it fail all 7 retries. Jan saw it "succeed", but clipboard get then returned data: VirtualBox clipboard sync wrote "Test" back.
      • The user's Windows VM has VirtualBox clipboard="bidirectional" (VBoxManage showvminfo), so it should reproduce Jan's variant.
      • The prototype compiles with x86_64-w64-mingw32-gcc -fsyntax-only against trunk headers.
    • Cause (hypothesis, needs a Windows run): win/tkWinClipboard.c.
      • Tk announces CF_UNICODETEXT with delayed rendering (SetClipboardData(CF_UNICODETEXT, NULL) in UpdateClipboard).
      • A clipboard listener (Windows clipboard history/cloud sync, VBoxTray, VNC, rdpclip) reacts to WM_CLIPBOARDUPDATE, opens the clipboard and calls GetClipboardData. That sends WM_RENDERFORMAT to Tk's thread and blocks with the clipboard open.
      • Tk's thread is still running the test script and does not process sent messages, so selection clear (UpdateClipboard(NULL), whose OpenClipboard fails silently) and the following clipboard get (OpenClipboard(NULL)) both fail. That produces "clipboard cannot be opened".
      • A retry that sleeps cannot help, because the holder is waiting for Tk itself. It works interactively because the event loop runs between commands.
      • When a synchronizer writes the text back (VirtualBox), the expected "doesn't exist" error cannot be produced at all, so clipboard-4.x/winClipboard tests also need a constraint or must tolerate that case.
    • Prototype ee397e0a55.patch (75 lines, against trunk, untested). OpenClipboardWait() retries for up to 500 ms and handles only *sent* messages between attempts (MsgWaitForMultipleObjects + PeekMessage PM_QS_SENDMESSAGE), so a pending WM_RENDERFORMAT gets answered. TkSelGetSelection and UpdateClipboard use it, and UpdateClipboard no longer calls EmptyClipboard/SetClipboardData after a failed open. It compiles (mingw -fsyntax-only) but has not been run.
    • Suggested ticket comment Not fixed: bug-ee397e0a55 was never merged, so the "Fixed" resolution is wrong. The sleep-retry there can't work if the window holding the clipboard is blocked on a WM_RENDERFORMAT sent to Tk's own (busy) thread. Retrying while handling sent messages should cure the "cannot be opened" failures, but tests whose expected result is an empty clipboard can't pass while a clipboard synchronizer (VirtualBox/VNC/RDP) copies the data back.
    Prototype verified 2026-10-24, Windows VM
    • Status VERIFIED (statistically; the failure is intermittent)
    • Applies to main ported to the -p0 fossil format (the file had a/ b/ prefixes). The code is unchanged.
    • Repro before: winClipboard-1.3 failed with "clipboard cannot be opened, another application grabbed it" in 1 of 11 unpatched runs. A stress loop with 0–150 ms random pauses (clipboard append / update / selection get / testclipboard) failed 2 of 150 times. After: the same loop failed 0 of 300 times, and winClipboard.test passed 5/5.
    • Test none new (existing winClipboard-1.3 is the one that fails intermittently; it depends on the VirtualBox clipboard sync timing)
    • Test file run winClipboard.test 8/0, clipboard.test 42/0, select.test 111/0 (pre-existing failure without the patch: winClipboard-1.3, intermittent)
    • Patch changes format only (-p1 → -p0)
    ⬜dc18394b422020Combobox dropdown frozen or visually reducedNEEDS-WINDOWS (needs two monitors with different scaling; duplicate of a9ee44102b)

    Open ticket dc18394b42 · 02. Appearance · filed 2020-08-20

    Full verdict

    NEEDS-WINDOWS (needs two monitors with different scaling; duplicate of a9ee44102b)

    • Platform Windows
    • Tested the script only, on X11 (single screen, geometry OK).
    • Cause per-monitor DPI is not supported. Tk uses one scaling factor per process, so it cannot handle a popdown on a monitor with a different DPI.
    • Prototype none (per-monitor DPI awareness is a design issue). Repro: win/dc18394b42.tcl (posts the dropdown, checks its position and mapping, prints VERDICT).
    • Suggested ticket comment Duplicate of a9ee44102b (mixed-DPI multi-monitor on Windows). Tk has a single scaling per process.
    ⬜a9ee44102b2019Windows 10 multi-screen dpi scaling problemNEEDS-WINDOWS (needs two monitors set to different Windows scaling, e.g. 100% and 150%). Possible duplicate of dc18394b42, which already has a repr…

    Open ticket a9ee44102b · 68. Win Window Operations · filed 2019-09-12

    Full verdict

    NEEDS-WINDOWS (needs two monitors set to different Windows scaling, e.g. 100% and 150%). Possible duplicate of dc18394b42, which already has a repro at win/dc18394b42.tcl.

    • Platform Windows
    • Tested X11 is not affected, because this depends on per-monitor DPI virtualization. The repro script runs cleanly on X11 main.
    • Cause likely Windows' DPI virtualization of system-DPI-aware processes. Per chrstphrchvz, overrideredirect toplevels take the scaling state of another toplevel. A real fix needs per-monitor DPI awareness (per-toplevel tk scaling), which is a large redesign.
    • Prototype none (too large). Repro: win/a9ee44102b.tcl (interactive; asks whether the red marker covers the button and prints a VERDICT).
    • Suggested ticket comment Still needs testing on Windows 10/11 with two monitors at different scaling settings (not just different resolutions). Probably the same issue as dc18394b42.
    ⬜d66e6fabad2019Complex alpha photo images: red and blue are swapped on WindowsNEEDS-WINDOWS. The reported symptom should be gone on trunk/main since backout 86970d9896 (2019-07-13) of efab40f95b. fvogel confirmed correct …

    Open ticket d66e6fabad · 41. Photo Images · filed 2019-06-05

    Full verdict

    NEEDS-WINDOWS. The reported symptom should be gone on trunk/main since backout 86970d9896 (2019-07-13) of efab40f95b. fvogel confirmed correct rendering on trunk in 2023-12. Branch bug-d66e6fabad (last afc2508990, 2023-11-29) was never merged. It re-applies efab40f95b, and chrstphrchvz reports it shows a wrong (blue) label background for earthmenu.png. The underlying ticket 21525158b0 (XImage masks on Windows) is still open.

    • Platform Windows
    • Tested X11 main renders correctly (the repro script reports NOT REPRODUCED). Windows not tested.
    • Cause win/tkWinImage.c/tkWinDraw.c BlendComplexAlpha(): RGB masks disagree with the XImage visual masks once efab40f95b is applied.
    • Prototype none (the branch already exists). Repro: win/d66e6fabad.tcl (SVG + 50%-alpha red/blue photo next to canvas reference swatches; yes/no verdict).
    • Suggested ticket comment On main the red/blue swap no longer occurs, because efab40f95b was backed out. The ticket could be closed as fixed on main, with the unmerged branch left for 21525158b0.
    ⏳fec832b6bc2018Cross-compiled Tk dll can't find resourcesNEEDS-WINDOWS

    Open ticket fec832b6bc · 85. Win Build · filed 2018-06-17

    ⏳ Since triage (updated 2026-11-11)

    Status now: Pending

    • 💬 2026-09-27 serhiy.storchaka: Not reproducible with 8.6.18 on Windows 11: cross-compiled with MinGW-w64 on Linux (with its own tclsh, and with tk86.dll loaded into an MSVC-built tclsh86t), built natively with MinGW, and built with nmake as shared and as static. For static builds wish.rc includes tk_base.rc. If this still happens with a current version, please give the exact build commands.
    Full verdict

    NEEDS-WINDOWS

    • Platform build (Windows, mingw cross-compile)
    • Tested nothing on Linux (no mingw toolchain here, and the failure only shows at run time on Windows). No win/ build-file changes between core-8-6-7 and core-8-6-8 other than the version bump in configure.in (the TIP 477 changes were in makefile.vc/rules.vc only).
    • Cause probably tkWinX.c:Tk_GetHINSTANCE() falling back to GetModuleHandle(NULL), i.e. tclsh.exe, because DllMain → TkWinSetHINSTANCE() never ran in that DLL. Then tkWinButton.c:InitBoxes() FindResourceW(module, "buttons") fails and panics. That points to the build setup (e.g. STATIC_BUILD, or a DLL mixed into a foreign MSVC binary distribution) rather than to the source. The later comment about makefile.vc concerns a different build (MSVC).
    • Prototype none (not reproducible here, and likely a build/config issue). Repro: win/fec832b6bc.tcl (run it with the tclsh of the tested build).
    • Suggested ticket comment Nothing in the configure/Makefile.in build changed between 8.6.7 and 8.6.8. The panic means Tk's HINSTANCE was not set from DllMain, so resources were looked up in tclsh.exe. Can you still reproduce this with a current 8.6/9.0 mingw build that is not mixed into another binary distribution? Otherwise we will close it.
    ⬜21525158b02018On MS Windows XImage data and Tk_Visual() return wrong informationNEEDS-WINDOWS (likely NOT-A-BUG / Won't Fix)

    Open ticket 21525158b0 · 63. Tk_Win Functions · filed 2018-03-01

    Full verdict

    NEEDS-WINDOWS (likely NOT-A-BUG / Won't Fix)

    • Platform Windows
    • Tested code inspection only. On X11, main tip (9.1b1): canvas image colours are correct (repro script also run on X11).
    • Cause win/tkWinX.c:TkWinDisplayChanged sets the visual masks to COLORREF layout (red 0xff, blue 0xff0000) because Tk pixel values on Windows are COLORREFs, and all GDI drawing depends on that. win/tkWinImage.c:XGetImage returns DIB data (BGRX), and XCreateImage hard-codes bitmap_unit=8. Changing the visual masks would break pixel semantics everywhere. The only core consumer, generic/tkCanvas.c (canvas image, written by the reporter), already swaps red and blue under #ifdef _WIN32 and uses bits_per_pixel. canvas-23.* check the colours and pass in Windows CI.
    • Prototype none (changing it would redesign Windows pixel semantics; the only in-core user is already worked around).
    • Suggested ticket comment On Windows, Tk pixel values are COLORREFs, so the visual masks describe that layout and cannot change without breaking all drawing. $canvas image handles the BGRX XImage layout explicitly, and canvas-23.* check its colours. Suggest closing as Won't Fix.
    • Repro win/21525158b0.tcl (checks canvas image colours, Tk 8.7+).
    ⬜1dc430ad592016Cursor blink rate in [text] widgets doesn't match Windows settingsNEEDS-WINDOWS (RFE, still unimplemented on main)

    Open ticket 1dc430ad59 · 18. [text] · filed 2016-11-13

    Full verdict

    NEEDS-WINDOWS (RFE, still unimplemented on main)

    • Platform Windows
    • Tested code inspection. win/tkWinDefault.h still hard-codes DEF_*_INSERT_OFF_TIME "300"/ON "600" and nothing calls GetCaretBlinkTime(). ttk now reads -insertofftime/-insertontime from the style or the option db (ttkBlink.c), so one option-db entry would cover both Tk and ttk.
    • Cause feature not implemented.
    • Prototype none (Windows-only; can't build/test). Suggested fix: at Tk init on Windows, if GetCaretBlinkTime() returns INFINITE, option add *insertOffTime 0 widgetDefault. Otherwise set insertOnTime/insertOffTime to that value. Repro script win/1dc430ad59.tcl.
    • Suggested ticket comment Still not implemented. A simple approach is to seed the option database (*insertOffTime / *insertOnTime) from GetCaretBlinkTime() at startup. That would cover the Tk widgets and ttk, since ttkBlink.c honors those resources.
    ✅6c0d7aec672016unicode text inputNEEDS-WINDOWS (inconclusive: FAIL on main, but it is not verified that the synthetic input reached the window)

    Open ticket 6c0d7aec67 · 44. Generic Fonts · filed 2016-02-17

    ✅ Since triage (updated 2026-11-11)

    Status now: Closed (Fixed)

    Full verdict

    NEEDS-WINDOWS (inconclusive: FAIL on main, but it is not verified that the synthetic input reached the window)

    • Platform Windows
    • Tested code check only. Branch win-unicode-keyinput (04cfb8ed11) was merged to main in a444f9c1a5 (2026-09-22). GetTranslatedKey() now reads WM_CHAR with GetMessageW() in tkmain, tk9.0 and tk8.6 (the 2026-10-10 trees). The VK_PACKET handling from this ticket was already there.
    • Cause WM_CHAR messages were read with GetMessageA(), so surrogate pairs from VK_PACKET input were lost (they arrived as ??).
    • Prototype none (fix already merged). Windows check: win/6c0d7aec67.tcl. It uses PowerShell SendInput(KEYEVENTF_UNICODE) (the on-screen keyboard path, VK_PACKET) to type U+0151 and U+1F600 into an entry, and prints PASS/FAIL and a VERDICT line. It needs 64-bit PowerShell.
    • Suggested ticket comment The fix for a9bd2ab98b (branch win-unicode-keyinput, merged in a444f9c1a5 to main, 9.0 and 8.6) reads WM_CHAR as Unicode, so emoji typed on the on-screen keyboard now arrive as one character. This can be closed once someone confirms on Windows.
    Windows VM, run 6
    • Platform Windows
    • Verdict NEEDS-WINDOWS (inconclusive: FAIL on main, but it is not verified that the synthetic input reached the window)
    • Tested SendInput(KEYEVENTF_UNICODE) of U+0151 and U+1F600 through PowerShell (run under tclsh with Tk loaded, so PowerShell output works): the entry stayed empty for both. Mouse input through the same PowerShell path did work for 227980ffff, but keyboard input also needs the foreground window, which is not checked.
    • Prototype none
    • Suggested ticket comment Needs a manual check with the on-screen keyboard: the automated SendInput test typed nothing into the entry, which may be a focus problem of the test.
    ✅562c3a3da52015transient windows don't appear in WinXPNEEDS-WINDOWS (NOT-REPRODUCED on X11)

    Open ticket 562c3a3da5 · 68. Win Window Operations · filed 2015-04-07

    ✅ Since triage (updated 2026-11-11)

    Status now: Closed (Fixed)

    • 🔧 2026-09-27 check-in 90f2a85de3 on win-transient-deiconify, in main, 9.0: Fix 562c3a3da5: on Windows a transient withdrawn before its container was mapped could not be deiconified. TkWmMapWindow() returned early without creating the
    • 🔧 2026-10-24 check-in b56b1a051f on trunk, in main, 9.0: Fix for 562c3a3da5: transient windows don't appear on Windows
    • 🔧 2026-10-24 check-in 83373d40a0 on core-9-0-branch, in main, 9.0, 8.6: Fix for 562c3a3da5: transient windows don't appear on Windows
    • 🔧 2026-10-24 check-in 8fcd84d452 on core-8-6-branch, in 8.6: Fix for 562c3a3da5: transient windows don't appear on Windows
    • 💬 2026-09-27 serhiy.storchaka: Still reproducible on Windows 11 with 8.6, 9.0 and 9.1, with the script from the description (still available at the link). The command renaming does not matter: a transient withdrawn before its container is mapped cannot be deiconified. [code] TkWmMapWindow() returned early for a transient whose container was not mapped, without creating the wrapper, so the window stayed never mapped and wm deiconify only recorde…
    • 💬 2026-10-24 nemethi: Serhiy, many thanks! Merged into main, core-9-0-branch, and core-8-6-branch by commits b56b1a05, 83373d40, and 8fcd84d4.
    Full verdict

    NEEDS-WINDOWS (NOT-REPRODUCED on X11)

    • Platform Windows
    • Tested main, 8.6 tip on Xvfb without a WM. Three reconstructed variants: a withdrawn transient deiconified from <Double-1>, a transient created in the handler, and a transient declared before the master is mapped. The transient was mapped in all three. The original script (wooledge.org URL) is not available offline, so this reconstruction is a guess.
    • Cause unknown. The report is 8.6.4 built with MinGW on WinXP, and the reporter patched out shobjidl.h.
    • Prototype none. Repro script: win/562c3a3da5.tcl (prints VERDICT and writes 562c3a3da5.out).
    • Suggested ticket comment Could not reproduce with a reconstruction of the described scenario on current Tk. The original script link is dead; please attach the script to the ticket if the problem persists with 8.6.17+/9.0 on a supported Windows version. Otherwise suggest closing as out of date.
    ⬜c95d4691ff2015winfo screenheight w/dual monitors returns the screen height of the primary displayNEEDS-WINDOWS (the VM has one monitor)

    Open ticket c95d4691ff · 68. Win Window Operations · filed 2015-01-11

    Full verdict

    NEEDS-WINDOWS (the VM has one monitor)

    • Platform Windows (also design-level on X11)
    • Tested Linux shows the same single-screen model: Xinerama/RandR monitors are one X screen, so screenwidth is the whole screen (see 985607f240). On Windows, win/tkWinX.c:TkWinDisplayChanged still uses GetDeviceCaps on the desktop DC, which returns the primary monitor.
    • Cause TkWinDisplayChanged fills the single Screen from GetDeviceCaps(desktop DC), which is the primary monitor.
    • Prototype none (a semantic change touching all WidthOfScreen users, which needs a TIP). Repro script: win/c95d4691ff.tcl. It lists monitors via PowerShell, puts a toplevel on each, compares winfo screenwidth/height with that monitor, and prints a verdict (SKIP with fewer than two monitors).
    • Suggested ticket comment Still the case in main. Tk keeps one screen per display, and making winfo screen* per monitor would change the meaning for all platforms and every user of the screen size. I propose closing this as Won't Fix and tracking per-monitor information as a TIP or RFE, together with the related tickets listed above.
    Windows VM, run 6
    • Platform Windows
    • Verdict NEEDS-WINDOWS (the VM has one monitor)
    • Tested PowerShell reports one monitor 1920x1200, so the script skipped.
    • Prototype none
    • Suggested ticket comment Needs a machine with two monitors of different sizes.
    ⬜82aadad30f2014Trackpad scrolling in doesn't work on some laptopsNEEDS-WINDOWS (and needs specific hardware: a Lenovo T420 UltraNav/Synaptics trackpad)

    Open ticket 82aadad30f · 68. Win Window Operations · filed 2014-12-10

    Full verdict

    NEEDS-WINDOWS (and needs specific hardware: a Lenovo T420 UltraNav/Synaptics trackpad)

    • Platform Windows
    • Tested none. Main's win/tkWinX.c now has trackpad heuristics for WM_MOUSEWHEEL/WM_MOUSEHWHEEL with sub-120 deltas, and high-resolution wheel support has been in 8.6 since around 8.6.8.
    • Cause unknown. Old Lenovo/Synaptics drivers send legacy WM_VSCROLL messages, or none at all, to windows without native scrollbars. If so, it's a driver issue that Tk can't see.
    • Prototype none. Repro script: win/82aadad30f.tcl (interactive, 20 s; logs the events received and prints VERDICT).
    • Suggested ticket comment Windows wheel handling has changed a lot since 8.6.3: high-resolution deltas, and a trackpad heuristic in 9.x. Please retest with a current release. Without a report from the affected hardware, suggest closing as out of date.
    ⬜2919709fff2009Windows menus don't honor Windows theme changesNEEDS-WINDOWS (very likely still present: code unchanged)

    Open ticket 2919709fff · 13. Win Menus · filed 2009-12-23

    Full verdict

    NEEDS-WINDOWS (very likely still present: code unchanged)

    • Platform Windows
    • Tested source inspection only. win/tkWinWm.c WM_SYSCOLORCHANGE is still an empty "XXX" case; win/tkWinColor.c stores GetSysColor() into the XColor pixel once at allocation (TkpGetColor), and "SystemMenu" colors are cached/refcounted by name, so menus keep the old color. tkWinMenu.c WM_SETTINGCHANGE only reloads SPI_SETNONCLIENTMETRICS/keyboard cues via SetDefaults().
    • Repro win/2919709fff.tcl (menubar with File menu; polls a fresh GetSysColor lookup of SystemMenu via a unique letter case of the name and compares with the color Tk holds for the menu background; prints REPRODUCED/NOT REPRODUCED after switching a contrast theme on Windows 11).
    • Cause system colors are resolved to fixed pixels at allocation; nothing refreshes them (or menu GCs) on WM_SYSCOLORCHANGE.
    • Prototype none (proper fix = on WM_SYSCOLORCHANGE refresh pixel of every cached WinColor with index >= 0 and reconfigure widgets/menus — a broader design change, >60 lines)
    • Suggested ticket comment Still present in the code: WM_SYSCOLORCHANGE is not handled, and System* colors are resolved to fixed pixels when allocated, so menus (and other widgets) keep the old colors until restart.
    ⏳1883850fff2008xptheme appearance on VistaNEEDS-WINDOWS. This is partly fixed: the vista theme was added in 8.6 and 8.5.8, and vistaTheme.tcl now has vsapi Combobox/Spinbox elements. Per pa…

    Open ticket 1883850fff · 88. Themed Tk · filed 2008-01-31

    ⏳ Since triage (updated 2026-11-11)

    Status now: Pending (Out of Date)

    • 💬 2026-09-27 serhiy.storchaka: The vista theme, added in 8.5.8 and 8.6 and used by default on Vista and later, addressed this. On Windows 11 with 9.1, the readonly and editable comboboxes and the spinbox look like the native controls. The remaining differences mentioned below (fuzzy ends of the indeterminate progressbar, translucent selection, hover animation) are Vista/7 visuals which are no longer used by current Windows versions.
    Full verdict

    NEEDS-WINDOWS. This is partly fixed: the vista theme was added in 8.6 and 8.5.8, and vistaTheme.tcl now has vsapi Combobox/Spinbox elements. Per patthoyts (2009), the readonly combobox look, the fuzzy marquee and the hover animation were still missing.

    • Platform Windows
    • Tested none on Linux (source inspection of library/ttk/vistaTheme.tcl)
    • Cause The ttk vista theme approximates the look through vsapi elements. Some Vista/7+ visuals (animated transitions, the readonly combobox drop-down-button style, the marquee progress bar) are not mapped.
    • Prototype none (visual/theme work, Windows only). Visual check script: win/1883850fff.tcl (shows each reported widget with a numbered label for comparison against native controls).
    • Suggested ticket comment The vista theme (8.5.8+) addressed most of this. The remaining items need re-checking on Windows 10/11 with current Tk: readonly combobox style, indeterminate progressbar and hover animation. If they are still wanted, this is better tracked as an RFE.
    ⬜1449550fff2006access violation crash in Tk_AttachHWNDNEEDS-WINDOWS (no deterministic repro; intermittent after long idle under load)

    Open ticket 1449550fff · 13. Win Menus · filed 2006-03-14

    Full verdict

    NEEDS-WINDOWS (no deterministic repro; intermittent after long idle under load)

    • Platform Windows
    • Tested code inspection only. Tk_AttachHWND (win/tkWinWindow.c) and MenuSelectEvent (win/tkWinMenu.c) are essentially unchanged. twdPtr == 0x1 means Tk_WindowId() of the menu returned garbage, which points to a stale or freed TkWindow in the RecursivelyClearActiveMenu chain rather than to Tk_AttachHWND itself. The second reporter's trace turned out to be unrelated (Remote Desktop colour depth, bug 1123042). Repro script: win/1449550fff.tcl (manual stress test that rebuilds cascades while you browse the menus; writes a verdict to 1449550fff.out; checked on X11 for Tcl errors only).
    • Cause unknown; possibly a menu destroyed while still referenced by the Windows active-menu chain
    • Prototype none
    • Suggested ticket comment No reproducer, and nobody has reported it against 8.6/9.x. Please reopen with a script or a current stack trace if it still happens.
    ⬜1123042fff2005Images are blank via Remote DesktopNEEDS-WINDOWS (likely hard to reproduce: current mstsc offers no 256-colour mode, 15/16-bit at least)

    Open ticket 1123042fff · 42. Photo Image|GIF · filed 2005-02-15

    Full verdict

    NEEDS-WINDOWS (likely hard to reproduce: current mstsc offers no 256-colour mode, 15/16-bit at least)

    • Platform Windows
    • Tested not tested (needs an RDP session with a low or changing colour depth)
    • Cause unknown. Probably the palette and colour-cache code in win/tkWinColor.c and the photo code assume the screen depth that was in effect at startup, which is wrong after an RDP depth change.
    • Prototype none. Repro: win/1123042fff.tcl (shows the earth/teapot photos and prints the screen depth. Reconnect with another colour depth, then press "Close images" to check for the "Tried to free a color" panic.)
    • Suggested ticket comment Can anyone still reproduce this on Windows 10/11? 256-colour RDP is no longer offered. Please try 16-bit, and reconnecting with a different depth while wish is running.
    ⬜870844ffff2004WinTk: many menus in menubar won’t popupNEEDS-WINDOWS (inconclusive)

    Open ticket 870844ffff · 13. Win Menus · filed 2004-01-05

    Full verdict

    NEEDS-WINDOWS (inconclusive)

    • Platform Windows
    • Tested nothing on X11 (Win32 resource limit). No check-ins mention the ticket.
    • Cause win/tkWinMenu.c. Every Tk menu owns a Win32 HMENU, and the menubar clone doubles that. Windows allocates about one more handle per menubar menu while one is open, so 3905 menus go past the 10000-handle quota.
    • Prototype none. A fix would avoid creating Win32 menus for the non-displayed originals. That is a change to the clone design, too big for a prototype.
    • Repro win/870844ffff.tcl. It builds the ticket's menubar, reads the USER handle count with GetGuiResources (through powershell), posts a menu, makes a new toplevel, and prints PASS/FAIL.
    • Suggested ticket comment This is the Windows USER-handle quota (10000 per process), the same limit as 31085d2747. It can be raised with USERProcessHandleQuota. Halving the cost would mean not creating Win32 menus for the original (non-clone) menus. Suggest closing as a platform limitation, or keeping it open as a low-priority RFE.
    Windows VM, run 6
    • Platform Windows
    • Verdict NEEDS-WINDOWS (inconclusive)
    • Tested The 3906-menu menubar was built, but the USER handle count needs PowerShell (which returns nothing under tktest) and the menu posting timed out in the Win32 menu loop. Under tclsh with Tk loaded the menus fail with "Failed to register menu window class".
    • Prototype none
    • Suggested ticket comment Inconclusive in the automated run; serhiy reproduced it by hand on 2026-09-21.
    ⬜846141ffff2003winSend.test failures and tkWinSend.c bugNEEDS-WINDOWS (send is compiled out of standard builds; not testable)

    Open ticket 846141ffff · 58. [send] · filed 2003-11-20

    Full verdict

    NEEDS-WINDOWS (send is compiled out of standard builds; not testable)

    • Platform Windows
    • Tested code check plus the X11 reference behaviour:
      • Main/9.0 X11 send -bogus x y gives bad option "-bogus": must be -async, -displayof, or --. That is X11 code since 9.0 (return TCL_ERROR in unix/tkUnixSend.c).
      • 8.6 X11 itself gives no application named "-bogus" (its loop also does break).
      • win/tkWinSend.c:349 still does break, so the Windows bad-option bug only matters against 9.x behaviour.
    • Cause win/tkWinSend.c Tk_SendObjCmd():
      • Option-parse break instead of return TCL_ERROR (9.x).
      • Remote errorInfo/errorCode are built wrongly.
      • winfo interps and the "target application died" message differ from X11.
      • winSend-6.x/9.x/10.x test the old DDE implementation.
    • Prototype none. The code is compiled out, and without Windows it can't be verified. The option-parse part is a one-liner: return TCL_ERROR as in tkUnixSend.c.
    • Windows check win/846141ffff.tcl. It reports SKIP when send does not exist (normal builds). Otherwise it checks option errors, winfo interps, and errorCode/errorInfo of a remote error against a second wish, with expectations matching X11 for the running version. Prints PASS/FAIL/VERDICT.
    • Suggested ticket comment The option-parsing part should simply return TCL_ERROR like tkUnixSend.c does since 9.0. winSend-6.x/9.x/10.x test the old DDE implementation and should be removed. The other failures need a decision on whether the COM send is still worth keeping.
    Windows VM, run 6
    • Platform Windows
    • Verdict NEEDS-WINDOWS (send is compiled out of standard builds; not testable)
    • Tested The send command does not exist in the makefile.vc build, so the script reported SKIP. The code findings from the Linux triage stand.
    • Prototype none
    • Suggested ticket comment Not testable with standard builds, since send is compiled out on Windows. See the code notes above.
    ⏳760693ffff2003menubuttons lockupNEEDS-WINDOWS (inconclusive)

    Open ticket 760693ffff · 13. Win Menus · filed 2003-06-25

    ⏳ Since triage (updated 2026-11-11)

    Status now: Pending

      Full verdict

      NEEDS-WINDOWS (inconclusive)

      • Platform Windows
      • Tested X11 8.6.18: 30 post/unpost cycles over 3 menubuttons, then no grab was left and the button still got Tk-level clicks. The X11 test says nothing about the Win32 mouse-capture problem in the report.
      • Cause unknown, probably the Win98 menu modal loop or mouse capture.
      • Prototype none (cannot reproduce)
      • Repro win/760693ffff.tcl. It posts and unposts menubuttons 30 times, then clicks a button with an OS-level click through powershell mouse_event, and prints PASS/FAIL.
      • Suggested ticket comment Not reproducible on current Windows with 8.6.18 and 9.1. The report is for Windows 98 and Tk 8.4.0. Closing as out of date.
      Windows VM, run 6
      • Platform Windows
      • Verdict NEEDS-WINDOWS (inconclusive)
      • Tested The script reaches the menu posting but its own watchdog fires while the menus are posted (the Win32 menu loop blocks the Tcl event loop), and the OS-level click through PowerShell is not available under tktest.
      • Prototype none
      • Suggested ticket comment The automated check was inconclusive. Not reproduced manually on Windows 11 earlier (2026-09-21); suggest closing as out of date.
      ⬜664111ffff2003tk_getOpenFile on WNT does not see VFSNEEDS-WINDOWS (by design: the native dialog only sees native filesystems)

      Open ticket 664111ffff · 37. [tk_get*File] · filed 2003-01-08

      Full verdict

      NEEDS-WINDOWS (by design: the native dialog only sees native filesystems)

      • Platform Windows
      • Tested X11 main: the Tk script dialog browses //zipfs:/lib/tk/tk_library fine (39 entries).
      • Cause win/tkWinDialog.c uses the native IFileDialog/GetOpenFileName, which knows nothing about Tcl VFS.
      • Repro win/664111ffff.tcl (opens the dialog on Tk's own zipfs library)
      • Prototype none. A possible improvement is for tk_getOpenFile/tk_getSaveFile/tk_chooseDirectory on Windows to fall back to the script dialog (::tk::dialog::file::) when [file system $initialdir] isn't native. That's an RFE.
      • Suggested ticket comment The native Windows dialog can't show Tcl VFS (zipfs, tclvfs). The Unix script dialog handles VFS fine. A fallback to the script dialog for non-native -initialdir would be a feature request.

      Cannot test 128

      🟢35a71bf4292026Aqua: imgPhInstance tests fail with default color profile for displayCANNOT-TEST (macOS)

      Open ticket 35a71bf429 · 41. Photo Images · filed 2026-10-18

      🟢 Since triage (updated 2026-11-11)

      Status now: Open

      • 🔧 2026-10-18 check-in 8316cf936d on aqua-photo-gamma-palette, in main, 9.0: Fix de6ef8276c: photo -gamma and -palette had no effect on macOS. Apply the fix for 32f2e19de8 to all platforms which render RGBA images, not only Windows:
      • 💬 2026-10-18 marc_culler: I don't know of any way to change the color profile in a Tcl script, but I think it would be possible to set the color profile in TkTest_Init with some platform specific code. I think that setting the display color profile at the script level is probably a missing feature in Tk, though.
      • 💬 2026-10-18 serhiy.storchaka: Instead of setting a color profile, the tests could compare a pixel of the image with a reference pixel drawn by Tk with the expected color, for example a canvas rectangle filled with #ff8000 next to the image, instead of with the color itself. Both go through the same conversion, so the result does not depend on the profile. For imgPhInstance-1.2 and the transparent cells in 2.1 the reference is a pixel of the ca…
      • 💬 2026-10-18 chrstphrchvz: To be clear, I only changed the display color profile in Settings as a temporary workaround; I am not suggesting that Tk try to change this setting, but rather only the colorspace object used in resetTkLayerBitmapContext (tkMacOSXWindowEvent.c).
      • 💬 2026-11-05 chrstphrchvz: Please see branch [bug-35a71bf4](timeline?r=bug-35a71bf4) (no need to do so before 9.0.5 is released, though). I have tested the proposed changes with a display set to use a non-sRGB color profile in Settings. I believe the bare-minimum fix for this issue would be to have TkpReadPixelSRGB() retrieve the colorspace a toplevel is using (rather trying to recreate it by duplicating whatever approach is currently in …
      Full verdict

      CANNOT-TEST (macOS)

      • Platform Aqua
      • Tested none. It depends on the display colour profile on macOS.
      • Cause the tests compare pixel values against fixed colours. On Aqua the layer bitmap context (resetTkLayerBitmapContext in macosx/tkMacOSXWindowEvent.c) uses the display colour space, so the values depend on the profile.
      • Prototype none. Possible fixes: compare against a reference pixel drawn by Tk (serhiy's suggestion), or use sRGB for the layer context. imgPhInstance-7.1 is only on branch aqua-photo-gamma-palette.
      • Suggested ticket comment Proposed fix: compare the image pixels with a reference rectangle filled with the expected colour in the same canvas, so both go through the same colour conversion. The blend tests stay notAqua. Marc or Christopher, would using sRGB in resetTkLayerBitmapContext be acceptable instead?
      🟢40dc19ee9e2026Regression: Canvas does not immediately display in Tk 9.1 on macOSCANNOT-TEST (macOS). Still open on main. Fix attempts are on unmerged branches:

      Open ticket 40dc19ee9e · 04. Canvas Basics · filed 2026-10-07

      🟢 Since triage (updated 2026-11-11)

      Status now: Open

      • 🔧 2026-10-08 check-in 1269dd2ba9 on bug-40dc19ee9e, in main, 9.0: Fix 40dc19ee9e: Canvas does not immediately display and f16cdb6d04: Aqua widgets not mapped immediately
      • 🔧 2026-10-21 check-in 9e4472ce01 on trunk, in main, 9.0: Disable event-9.1 testcase on Aqua, due to bug 40dc19ee9e
      • 💬 2026-10-08 tberg: Marc Culler wrote on the core mailing list 2026-10-08: ``` I was able to reproduce the issue on macOS 27 with Tk 9.1 when I put those commands into a script file and ran it with wish9.1. Also, if I add the line: update idletasks At the end of the script, I get a black canvas with a blue rectangle. This is one of the loss-lose situations that arise with the few (now 7) remaining situations where Aqua runs a "ne…
      • 💬 2026-10-08 tberg: I find it difficult to judge what the right "fix" is here. I agree that these nested event loop calls should never be needed. But what is the alternative? I think the users expect with such a script that the canvas will be visible right away. It is hard to explain why a user needs update idletasks here, especially since this does not affect other cases such as when just placing a ttk::button in a simple app. Doc…
      • 💬 2026-10-08 serhiy.storchaka: See also f16cdb6d04 (reported for 8.6.12): a listbox with a bad -yscrollcommand is not displayed and its error is not reported until an event or update idletasks arrives. It may have the same cause; a fix here should be checked with that script too.
      • 💬 2026-10-08 marc_culler: Tk does not use the macOS event loop at all. It uses the AppKit event queue, however. The Tk checkproc removes all of the AppKit events from the AppKit queue and translates them to XEvents which are posted on the Tcl event queue. Tcl does not use the macOS event loop to process user events. It uses its own queue. However, as I learned recently when dealing with the macOS 27 menu issue, Tcl_DoOneEvent does create r…
      • 💬 2026-10-08 tberg: OK, thanks for the explanation! I am not too familiar with these details. I'm, however, still a bit puzzled. You say, Tk does not use the macOS event loop. In the next paragraph you say Tcl_DoOneEvent is using the macOS event loop. OK, that is a Tcl C API command and not a Tk command but still it is used by the Tk code. Anyway, let's assume it is really not a problem of the event loop as such. What is making some …
      • 💬 2026-10-08 marc_culler: An update after looking at the code and doing some debugging. First, I was wrong when I said that the nested event loop: while(Tcl_DoOneEvent(TCL_IDLE_EVENTS)){} is run once per Tk instance. It get run once in the lifetime of a toplevel window, on the first call to updateLayer for that window, before the layer's tkLayerBitMapContext is displayed for the first time. This is what the debugging shows: % wish9.1 torst…
      • 💬 2026-10-08 marc_culler: Tk only deals with user events - mouse, and keys. it does not use the AppKit event loop for that. It only repurposes the AppKit queue. Tcl uses the AppKit event loop to process file and socket events, which is independent of Tk.
      • 💬 2026-10-09 chrstphrchvz: The clientData for updateLayerIdleTask() should be cast to TKContentView* rather than TKWindow*: `` macosx/tkMacOSXWindowEvent.c:416:13: warning: 'TKWindow' may not respond to 'updateLayer' 416 | [winPtr updateLayer]; | ~~~~~~ ^ ` It is also important that updateLayerIdleTask() does not try to use a deallocated TKContentView and cause a crash. But, my impression of how Apple intended updateLayer` to be…
      • 💬 2026-10-09 tberg: Yes, that makes it work but I think what I wrote earlier also explains what you are saying. You said > Then, in the first call to updateLayer the nested event loop runs DisplayCanvas twice, along with 2 other idle tasks. However, for some reason, those two calls to DisplayCanvas do not seem to do any drawing. Neither the canvas background nor the rectangle actually get drawn in the backing store CGImage." I still …
      • 💬 2026-10-09 marc_culler: Christopher, without claiming that the idle task is the correct solution, I just want to point out that I don't think that the current updateLayer, breaks your rule: updateLayer is not supposed to do a lot of work, so that there is no problem with it being called often. To me, this means that if queued Tcl events need to be handled to perform drawing, then they should be handled before, rather than during, the cal…
      • 💬 2026-10-09 marc_culler: Torsten, I think your suggestion makes sense. And it still satisfies the condiition I mentioned in my previous comment: The code: if (! [self onScreen]) { //printf("Running event loop.\n"); while(Tcl_DoOneEvent(TCL_IDLE_EVENTS)){} [self setOnScreen:YES]; } Runs exactly once for each toplevel. I think I should rename "onScreen". That name is misleading since its value does not indicate whether the window is current…
      • 💬 2026-10-09 marc_culler: Torsten, I think your analysis is completely correct. I have removed the idle task hack and replaced it with your simple suggestion of not accessing the context until after the mini-event loop has run (in the unique case where it does get run, namely the first time that updateLayer is called for a new toplevel window.) This makes your script work correctly and it makes Christopher's two scripts from f16cdb6d04 w…
      • 💬 2026-10-09 tberg: Great, thanks for confirming. I wasn't totally sure here, so it is good to have it confirmed by another pair of independent eyes. I find this event code difficult to understand as it has so many facets. It took me quite some time to understand the relevant parts regarding this phenomenon.
      • 💬 2026-10-09 marc_culler: I have just pushed a different solution to this problem to the bugfix branch. It addresses Christopher's concern about doing too much work in updateLayer. There were two flaws in my argument that it is OK to process idle tasks in updateLayer because it only happens once for each toplevel. The first is that it was not true that it only happens once. It was also happening when a window is deiconified. The second fla…
      • 💬 2026-10-09 jan.nijtmans: So, it looks like we are getting somewhere. Great!
      • 💬 2026-10-10 chrstphrchvz: The approach in 6964ffb4 does seem better, although I do not fully understand why it is needed. I think the danger in deferredUpdateLayer() is not that (TKContentView *)clientData might be nil (self is never nil during updateLayer, right?), but rather that it might already be deallocated by the time deferredUpdateLayer() is called. I think Tcl_CancelIdleCall(deferredUpdateLayer, deadView) would need to be done f…
      • 💬 2026-10-10 marc_culler: > I think the danger in deferredUpdateLayer() is not that (TKContentView > *)clientData might be nil (self is never nil during updateLayer, right?) Right. Thanks. I think I know how to deal with that. I will take another stab at it.
      • 💬 2026-10-10 marc_culler: I think the guard in deferredUpdateLayer is now correct. I added a boolean property windowIsDead to the TKContentView class and I set that to YES in tkWmDeadWindow. I check the property in deferredUpdateLayer. I think this fix is now ready to merge, but I am going to ask Nicolas Bats to test against his app. His report of artifacts when the main window of his app is withdrawn, configured, and deiconified was the o…
      • 💬 2026-10-10 chrstphrchvz: TkWmDeadWindow() releases the dead NSWindow, which in turn can release its contentView (i.e. deadView, because the NSWindow holds a strong reference to it). If deferredUpdateLayer() is called afterward and view was deallocated, then trying to check windowIsDead can crash. If the idle event loop in deferrredUpdateLayer() handles drawing idle tasks, they will call TkMacOSXRestoreDrawingContext(), and so TkMacOSXUpda…
      • 💬 2026-10-10 marc_culler: Nicolas reports that he sees the artifacts with the bugfix branch. Specifically, his root window appears on the screen at the correct size and then, after a short but noticeable delay, all the widgets appear at once. I think this means that the event loop and the idle task are useless. They are not needed to fix the issue in this ticket. But I am at a loss to explain why the previous incarnation of the event loop …
      • 💬 2026-10-10 marc_culler: I have now abandoned the inner event loop completely, which should be a relief to all of us, and have created a new bugfix branch branch bug-40dc19ee9e-alt which takes a completely different approach. My free chatbot (Gemini) told me that updateLayer is never called for a View whose window is off scteen, except that there is a trick which will force it to be called on the view of an off screen window. The trick is…
      • 💬 2026-10-11 marc_culler: Nicolas reports that the new bugfix branch does not solve the problem with his app, and that the reproducer written by Claude reports OK when things are not OK, but a close examination of the result shows that the script did detect the error but did not recognize it as an error. It sounds like the issue is that the drawing is not being done while the window is withdrawn, so it doesn't help to call updateLayer whil…
      • 💬 2026-10-13 tberg: This is all really complex. I wonder how you can even grasp this ... very cool. But what is now the right solution? It seems to me the only fix working for all scenarios is the simple change to updateLayer that I proposed. Of course, I see that this is not "the right solution" but it fixes the immediate problem for now. Then, another ticket could be created for the deeper logic to work on. That seems de-coupled …
      • 💬 2026-10-16 marc_culler: Too complex. The tip_765 branch fixes this problem as well as Nicolas' problem and some other stuff too, like the failures of tests event-9.11 - event-9.20 and the zombie windows that appear when running the tests. The fix for this problem is just that it removes the event loop from the updateLayer method, which was only there to deal with the artifacts when a window is first shown. With tip_765 the window can be …
      • 💬 2026-10-17 chrstphrchvz: Marc, is there a script available for others to try with tip_765 which uses -hidden to avoid visible artifacts?
      • 💬 2026-10-20 marc_culler: Nicolas provided me with a script written by Claude which creates a window containing a canvas containing a specified number of items (6000 being a typical value but 60000 can be used as well). It withdraws the window, configures the canvas, then deiconifies it. It also loads a custon dynamic library containing code which monitors all calls to updateLayer (by overriding the method on load) and checks pixel colors …
      • 💬 2026-10-20 marc_culler: One thing that I did when testing tip_765 was to add a print statement in updateLayer which reports the visiblity state of the window. That allows checking whether drawing is happening for off screen windows.
      • 💬 2026-10-20 marc_culler: Nicolas gave me permission to attach his testing code, so I will attach a tarball to this ticket containing the files he sent to me plus my modified version of the main script.
      • 💬 2026-10-20 marc_culler: Unfortunately the test code does not produce visual artifact that are easy for me to see. It relies on the monitoring code in the dylib. (And, incidentally, to compile the dylib look in the source file - it contains a gcc command that can be adapted to build the library.) However, the visual artifacts are apparently clearly visibile with Nicolas' app. So I relied on his eye for the ultimate conclusion about whethe…
      • 💬 2026-10-26 chrstphrchvz: I have not reviewed tip_765 very thoroughly. I do observe the re-enabled event-9.2 test sometimes failing on a real computer (albeit one older than what macOS 15 Sequoia officially supports). Normally, when event-9.2 starts, the root toplevel iconifies and .top1 appears immediately; after pausing, .top2 then appears at a position overlapping .top1. What I sometimes observe is that .top1 does not appear until after…
      • 💬 2026-10-27 marc_culler: I have not seen failures of event-9.2, but I think I may have seen windows appearing in a different order than the one in which they were created. This seemed to occur in sporadic failures in the wm-stackorder tests. (Either they appeared on screen in the wrong order or they failed to move front after being mapped.) What could cause that? I understand that mapping windows to the screen and ordering them is handled…
      • 💬 2026-10-27 marc_culler: Thanks for pointing out the collision of WM_HIDDEN. I will change that right away.
      Full verdict

      CANNOT-TEST (macOS). Still open on main. Fix attempts are on unmerged branches:

      • Platform Aqua
      • Tested none (Aqua only)
      • Cause macosx/tkMacOSXWindowEvent.c updateLayer reads tkLayerBitmapContext before the nested idle loop added in 6323022932. That loop can resize the view and replace the context, so an empty image is shown.
      • Prototype none (macOS only; branches already exist)
      • Suggested ticket comment Still open on main, waiting for TIP 765 (branch tip_765) or Torsten's minimal change: read the context after the idle loop in updateLayer. If the TIP vote is far off, the minimal change could go into 9.1 first.
      ✅061f633f432026MacOS aqua: <<ContextMenu >> binding sometimes triggers for incorrect toplevelCANNOT-TEST (macOS)

      Open ticket 061f633f43 · (unused) · filed 2026-09-18

      ✅ Since triage (updated 2026-11-11)

      Status now: Closed (Fixed)

      • 🔧 2026-09-29 check-in db5c579a55 on aqua-right-click-target, in main, 9.0: Fix 061f633f43: MacOS aqua: <<ContextMenu>> binding sometimes triggers for incorrect toplevel. Clicking with the right or middle button does not make the wind
      • 🔧 2026-10-08 check-in 3362fcc006 on trunk, in main, 9.0: Fix 061f633f43: MacOS aqua: <<ContextMenu >> binding sometimes triggers for incorrect toplevel
      • 🔧 2026-10-08 check-in 60a0dc7756 on core-9-0-branch, in main, 9.0, 8.6: Fix 061f633f43: MacOS aqua: <<ContextMenu >> binding sometimes triggers for incorrect toplevel
      • 🔧 2026-10-08 check-in ba8c94c23f on core-8-6-branch, in main, 9.0, 8.6: Fix 061f633f43: MacOS aqua: <<ContextMenu >> binding sometimes triggers for incorrect toplevel
      • 💬 2026-09-29 serhiy.storchaka: A right or middle click does not make the window the key window, so tkEventTarget is still the toplevel which has the focus. The event is then sent to the widget of that toplevel at the coordinates of the clicked window. A left click resets the target to the clicked toplevel, but a right or middle click does not. The same code is in 8.6 and 9.0. Proposed fix in branch aqua-right-click-target: right and middle butt…
      • 💬 2026-10-08 jan.nijtmans: Fixed 3362fcc006, and in core-9-0-branch and core-8-6-branch. I tested it, it works fine. Many thanks!
      Full verdict

      CANNOT-TEST (macOS)

      • Platform Aqua
      • Tested none. On X11, button events always go to the window under the pointer, so the report is about Aqua mouse-event routing
      • Cause unknown. Probably in macosx/tkMacOSXMouseEvent.c: when you click in an inactive window, the target is chosen from the key window or focus instead of the window under the pointer.
      • Prototype none (Aqua only)
      • Suggested ticket comment Aqua-specific (X11 delivers the click to the window under the pointer). Needs someone with macOS to check the target selection in tkMacOSXMouseEvent.c for clicks into an inactive toplevel.
      ✅2ef5dd80362026Aqua: "pack forget" leaves window on screenCANNOT-TEST (macOS). X11 is not affected (main, 9.0, 8.6: the frame is unmapped and the root window gets an Expose event)

      Open ticket 2ef5dd8036 · 48. Geometry Management · filed 2026-07-30

      ✅ Since triage (updated 2026-11-11)

      Status now: Closed (Fixed)

      • 🔧 2026-09-29 check-in 1b1c0ed60e on aqua-unmap-redraw, in main, 9.0: Fix 2ef5dd8036 and 307b06cbc2: Aqua: "pack forget" leaves window on screen. Since 9.0, unmapping a non-toplevel window no longer redrew its toplevel, so the
      • 🔧 2026-10-06 check-in 19dc130ec3 on trunk, in main, 9.0: Fix 2ef5dd8036 and 307b06cbc2: Aqua: "pack forget" leaves window on screen
      • 🔧 2026-10-06 check-in e1e4cbd699 on core-9-0-branch, in main, 9.0: Fix 2ef5dd8036 and 307b06cbc2: Aqua: "pack forget" leaves window on screen
      • 💬 2026-09-29 serhiy.storchaka: Since 9.0, XUnmapWindow() does not redraw the toplevel when a non-toplevel window is unmapped, so the area from which the window was removed is not redrawn and no Expose events are generated. Since fe5604ad44, XMapWindow() and XDestroyWindow() schedule the redraw in an idle task, but XUnmapWindow() does not. The same cause was found in 307b06cbc2. Proposed fix in branch aqua-unmap-redraw: XUnmapWindow() schedu…
      • 💬 2026-10-06 jan.nijtmans: Fixed in 19dc130ec3, also backported to core-9-0-branch Many thanks!
      Full verdict

      CANNOT-TEST (macOS). X11 is not affected (main, 9.0, 8.6: the frame is unmapped and the root window gets an Expose event)

      • Platform Aqua
      • Tested main, 9.0, 8.6 on Xvfb, no WM
      • Cause likely macosx/tkMacOSXSubwindows.c:XUnmapWindow. For a non-toplevel, 8.6 invalidated the parent view region (TkMacOSXInvalidateViewRegion on the parent's visRgn + TkMacOSXUpdateClipRgn + addTkDirtyRect). In 9.0 and later only TkMacOSXInvalClipRgns(parent) is left, so nothing redraws the uncovered area and no Expose is generated. It was dropped on branch cgimage_with_crossing (commented out in fe5604ad44) and merged to trunk in 1dfe4fe90a, 2024-07-08.
      • Prototype none (cannot build or test Aqua). Fix direction: in the else branch, after TkMacOSXInvalClipRgns, invalidate the rectangle of the unmapped window in the parent's view (e.g. TkMacOSXInvalidateViewRegion with the window's bounds, or setNeedsDisplayInRect:), so the parent is redrawn and gets Expose.
      • Suggested ticket comment Probably a regression from the cgimage_with_crossing merge 1dfe4fe90a (2024-07-08). XUnmapWindow no longer invalidates the area of the unmapped subwindow in the parent's view; it only invalidates clip regions. X11 is not affected.
      ✅33b3af1cea2026macOS 27 beta: modal dialog opened synchronously from a menu command becomes unresponsiveCANNOT-TEST (macOS 27 beta only)

      Open ticket 33b3af1cea · 11. Aqua Menus · filed 2026-07-21

      ✅ Since triage (updated 2026-11-11)

      Status now: Closed (Fixed)

      • 🔧 2026-09-29 check-in a235ccfcbd on bug-33b3af1cea, in main, 9.0: Fix 33b3af1cea: on macOS 27 dialog windows cannot be launched by a menu item command
      • 🔧 2026-10-05 check-in 8f02268ec8 on core-9-0-branch, in main, 9.0, 8.6: Fix 33b3af1cea and CPython issue #158053: hang on macOS 27 when a menu command opens a dialog
      • 🔧 2026-10-05 check-in 1e9833bf6e on trunk, in main, 9.0: Fix 33b3af1cea and CPython issue #158053: hang on macOS 27 when a menu command opens a dialog
      • 🔧 2026-10-05 check-in b7248183b6 on core-8-6-branch, in main, 9.0, 8.6: Fix 33b3af1cea and CPython issue #158053: hang on macOS 27 when a menu command opens a dialog
      • 💬 2026-09-29 marc_culler: This is a copy of the comment I just uploaded to the Python github site about this issue: [Issue 1580530](https://github.com/python/cpython/issues/1580530) I have at least identified what changed between macOS 26 and 27. Of course, with Apple, we are dealing with a black box. So I can only report symptoms, not underlying causes. The background is: - Tk uses Apple's event loop as an event source, removing events fr…
      • 💬 2026-09-30 jan.nijtmans: This looks like a blocker for Tk 9.0.5. +1
      • 💬 2026-09-30 marc_culler: Yes, I think it is a blocker for 9.0.5. I have pushed a bugfix branch bug-33b3af1cea which uses the workaround suggested here. All menu actions are run as idle tasks instead of being called directtly. I am not comfortable with this workaround because I cannot explain why it works. However, as far as I can tell, it does work. I suspect the answer to why it works may be buried in the Tcl notifier for macosx. If that…
      • 💬 2026-10-05 marc_culler: Here is my current best guess. People claim that the menubar display has been completely changed in macOS 27 and now uses a private undocumented framework named MenuBarClientCore. Evidently that framework changed the behavior of the runloop which is started when a menu is opened. In macOS 26 a runloop is started in NSEventTrackingRunLoopMode which terminates when a menuitem is selected. The action for the menuitem…
      • 💬 2026-10-05 marc_culler: I have merged bug-33b3af1cea into 9.1, 9.0 and 8.6. I am closing this ticket now.
      • 💬 2026-10-11 chrstphrchvz: Reopening this if I may. The current fix passes _eventInterp to Tcl_DoWhenIdle() without preserving it, which leaves open the possibility of a use-after-free if _eventInterp is deleted before showAboutIdleTask() is called. The following tclsh script triggers the problem, at least after 8e32cb7e and 1ab7a5f2 are fixed first: interp create i interp eval i package require Tk interp eval i proc tkAboutDialog x {{p…
      • 💬 2026-10-11 marc_culler: Thank you. Of course you may. I did not see any problem when running that script, but I added the preserve call anyway. I released _eventInterp in the idle task itself. I assume that is what you had in mind. I just pushed that change to the tip of main.
      • 💬 2026-10-11 chrstphrchvz: Thanks. Preserving _eventInterp is indeed one option, although it might be desirable to also check Tcl_InterpDeleted() in showAboutIdleTask() to prevent an "attempt to call eval in deleted interpreter" error. Another option would be to use Tcl_CallWhenDeleted(). But with either approach, I am not sure which would be better: letting tkAboutDialog silently fail to run, or emitting some message explaining t…
      • 💬 2026-10-11 marc_culler: I think I would lean towards emitting a message. In a real app, it is hard for me to imagine how a user would be in the situation where an "About" dialog is being requested from a deleted interpreter. A user might open the dialog to check the version of the software or on the first run to find out whatever information is available. But that is not likely to involve a deleted interpreter. So the message would be ai…
      • 💬 2026-10-13 chrstphrchvz: I think even developers could find that error message confusing, whether or not their program uses eval/bgerror/etc. Several Tkinter users had encountered 310c74ec, for example, and it seemed not uncommon for such issues to prompt them to believe (or be told) that their code was responsible for the error, but then not figure out exactly why.
      • 💬 2026-10-13 chrstphrchvz: There are also the options to document the behavior (e.g. "If the main interpreter is deleted, tkAboutDialog will not be called"); or, match the existing fallback behavior in -[TKApplication orderFromStandardAboutPanel:], which is to show the standard about dialog (the following works, but might not be ideal): ``` --- macosx/tkMacOSXMenus.c +++ macosx/tkMacOSXMenus.c @@ -23,7 +23,12 @@ static void ShowAboutIdleTas…
      • 💬 2026-10-16 marc_culler: I think this is sufficient, so I will close the ticket.
      • 💬 2026-10-16 chrstphrchvz: Can 1bc92655 please be backported to 8.6/9.0? Thanks
      Full verdict

      CANNOT-TEST (macOS 27 beta only)

      • Platform Aqua
      • Tested none
      • Cause per marc_culler, the switch from NSEventTrackingRunLoopMode (menu) to a modal runloop mode apparently broke in macOS 27 beta. It may be an Apple bug.
      • Prototype none
      • Suggested ticket comment Wait for the macOS 27 release. If it persists, run menu commands (or the dialog calls) from an idle callback on Aqua, as the reporter's workaround does.
      💬b964b98d1b2026ttk manager relayout use-after-free during teardown (Aqua)CANNOT-TEST (macOS, and no script). NOT-REPRODUCED on X11

      Open ticket b964b98d1b · (unused) · filed 2026-07-04

      💬 Since triage (updated 2026-11-11)

      Status now: Open

      • 💬 2026-10-04 anonymous: Yes, sorry for the long silence. (I should register somewhere so I get notigication, I suppose.) The attached b964b98d-repro.tcl is a standalone script for plain wish: wish9.1 b964b98d-repro.tcl ?iterations? It is not Aqua-only; the description's "not reproduced on X11" was wrong for this reduced form. A ttk::panedwindow has four panes, and a binding on one pane destroys that pane. Tk_MoveResizeWindow delivers syn…
      • 💬 2026-10-07 marc_culler: Thanks! I will take a look.
      Full verdict

      CANNOT-TEST (macOS, and no script). NOT-REPRODUCED on X11

      • Platform Aqua
      • Tested main (--enable-symbols) under valgrind on Xvfb: 20 cycles of a nested ttk::panedwindow (depth 3) plus a ttk::notebook, with update in <Destroy> bindings of the leaves so the idle relayout runs in the middle of teardown. No errors.
      • Cause not confirmed. The mechanism in the report does not match the code. Tk_DestroyWindow always sends DestroyNotify before it frees the TkWindow, and ContentLostEventHandler removes the content from mgr->content right then. So a freed content window should never be left in the list, even if an idle runs during teardown. The only gap is that a content window which is TK_ALREADY_DEAD but not yet removed can still be mapped (Tk_MapWindow on a half-dead window). That window is still valid memory.
      • Prototype none
      • Suggested ticket comment Could not reproduce on X11, even when the pending relayout is forced to run in the middle of teardown. Content is removed from the manager on DestroyNotify, before the window is freed. A Tcl script that crashes on Aqua is needed.
      ⬜637e74c1e72026TKApplication unrecognized selector causes program to abortCANNOT-TEST (macOS; the maintainers consider it a design decision / won't fix)

      Open ticket 637e74c1e7 · 83. Mac OS X Build · filed 2026-06-04

      Full verdict

      CANNOT-TEST (macOS; the maintainers consider it a design decision / won't fix)

      • Platform Aqua
      • Tested none
      • Cause macosx: Tk assumes NSApp is its TKApplication subclass. If another toolkit (GTK via PyGObject) creates NSApp first, [NSApp macOSVersion] etc. raise an unrecognized-selector exception. A fix would move TKApplication state into categories or a delegate, which is a redesign. Culler and Walzer declined, and Jan asked the reporter for a concrete patch on 2026-08-19; there has been no reply.
      • Prototype none (macOS; redesign)
      • Suggested ticket comment No concrete patch has been proposed in the month since Jan's request. Suggest closing as "Won't fix" unless one appears. A possible stopgap is to fail with a clear Tcl error instead of aborting when [NSApp isKindOfClass:[TKApplication class]] is false at Tk init.
      🔧e898fa74c52026Aqua: tests send-13.1 and send-13.2 fail when invoked with tcltest option "-singleproc 0"CANNOT-TEST (macOS). The analysis suggests a test-suite issue, not a bug in send.

      Open ticket e898fa74c5 · 58. [send] · filed 2026-05-07

      🔧 Since triage (updated 2026-11-11)

      Status now: Open

      • 🔧 2026-10-11 check-in 5a50c20986 on send-tests-other-apps, not merged: Fix e898fa74c5: send-13.1 and send-13.2 failed on Aqua with -singleproc 0, because [winfo interps] also listed the parent test process. Check only the names o
      • 💬 2026-10-11 serhiy.storchaka: The extra "all.tcl" is the parent test process, which is itself a Tk application with -singleproc 0. On X11 it is hidden from send-13.1 and send-13.2 because send.test overwrites the registry property at the top level (testsend prop root InterpRegistry); without that, the same failures occur on X11. On Aqua the registry is the file /tmp/TkAppnames, which testsend does not touch, so [winfo interps] still lists the …
      • 💬 2026-10-11 erikleunissen: I confirm that tests send-13.1 and send-13.2 on branch send-tests-other-apps both PASS under macOS/aqua and Linux/x11, using the invocation with "-singleproc 0".
      • 💬 2026-10-11 erikleunissen: B.t.w. I don't understand the purpose/workings of these tests (both old and new versions) enough to express a judgment w.r.t. fitness for merging into the main development branches.
      Full verdict

      CANNOT-TEST (macOS). The analysis suggests a test-suite issue, not a bug in send.

      • Platform Aqua (test suite)
      • Tested X11 main with -singleproc 0: send-13.x pass there, because on X11 earlier tests overwrite the root InterpRegistry property, which drops the parent "all.tcl" app. On Aqua, winfo interps lists running Tk processes, so the parent all.tcl process shows up. Side finding: on X11, send-8.16 fails ("target application died") whenever any other Tk app is running on the display, including the all.tcl parent under -singleproc 0.
      • Cause tests/send.test 13.1/13.2 expect exact winfo interps results, which assumes no other Tk app is running.
      • Prototype none (macOS). Suggested fix: in 13.1/13.2, remove the apps that existed before the test (winfo interps taken in setup) from the result, or constrain the tests to singleproc-only runs.
      • Suggested ticket comment The extra "all.tcl" entry is the parent test-runner process, which is itself a Tk app when -singleproc 0 is used. The tests should ignore apps that were already running before the test.
      ⬜4a2070f0d32026"wm iconbitmap" hangs on macOS 26 IntelCANNOT-TEST (macOS 26 Intel only; no hardware available to the maintainers either)

      Open ticket 4a2070f0d3 · (unused) · filed 2026-04-02

      Full verdict

      CANNOT-TEST (macOS 26 Intel only; no hardware available to the maintainers either)

      • Platform Aqua
      • Tested X11 sanity check: wm iconbitmap . hourglass finishes in under 0.2 s on main and 8.6.
      • Cause unknown. It is probably in macosx TkpGetNativeAppBitmap / the NSImage creation for the built-in bitmap (for example an NSWorkspace icon lookup that is slow on Tahoe Intel).
      • Prototype none
      • Suggested ticket comment Can't reproduce without macOS 26 Intel hardware. A sample wish stack while it hangs would show where it blocks.
      ⬜591829e9482026Incorrect initial pointer window under macOS/aquaCANNOT-TEST (macOS)

      Open ticket 591829e948 · 99. Other · filed 2026-02-15

      Full verdict

      CANNOT-TEST (macOS)

      • Platform Aqua
      • Tested none. On X11 the reporter's expected behavior already holds, per the ticket.
      • Cause two issues. Issue B (FrontWindowAtLocation counted unmapped windows) was fixed and merged by erikleunissen into trunk, 9.0 and 8.6 on 2026-03-14. Issue A: macosx/tkMacOSXInit.c:TkpInit runs idle handlers early, so . is mapped synchronously and wm ismapped . returns 1. Branch bug-591829e948-A (last rebased 2026-08-19, 1bae257f8c) proposes deferring that mapping. It needs review by someone who knows macOS.
      • Prototype none (macOS; a fix branch already exists)
      • Suggested ticket comment Part B is fixed on all branches. Part A is waiting for a macOS review of branch bug-591829e948-A.
      💬5b06d8b94a2026Wrong RGB values for systemDisabledControlTextColorCANNOT-TEST (macOS)

      Open ticket 5b06d8b94a · 66. Aqua Window Operations · filed 2026-01-15

      💬 Since triage (updated 2026-11-11)

      Status now: Open

      • 💬 2026-09-29 serhiy.storchaka: Since macOS 10.14 disabledControlTextColor and some other semantic colors are not opaque. GetRGBA() keeps their alpha: * When drawing, a background in such a color is translucent black over the window background. It looks like #b2b2b2 at first, but in 9.x every redraw adds another layer, so it gets darker after the window is moved. * TkpGetColor() ignores alpha, so winfo rgb returns 0 0 0. Tk colors are opaque, so…
      Full verdict

      CANNOT-TEST (macOS)

      • Platform Aqua
      • Tested none
      • Cause (from reading the code): macosx/tkMacOSXColor.c:TkpGetColor handles semantic colors by taking rgba[0..2] from GetRGBA and ignoring rgba[3]. On 10.14+, disabledControlTextColor is black with low alpha. It is drawn as gray after compositing, but winfo rgb returns 0 0 0. A fix could blend against the window background (or the appearance's background) when alpha < 1 before filling the XColor. The 9.x color change when the window moves is probably the same alpha composited over whatever lies beneath.
      • Prototype none (macOS)
      • Suggested ticket comment The XColor for semantic colors ignores the alpha component (tkMacOSXColor.c, TkpGetColor). Since 10.14 this color is translucent black, which is why winfo rgb gives 0 0 0. Blending with the window background when alpha < 1 would give the displayed value.
      ✅aeb2ed51ab2025Aqua: Unretained NSWindow reference passed to showRootWindow() idle taskpatchCANNOT-TEST (macOS). Code check: the bug is still present in main, 9.1 and 9.0 (Tcl_DoWhenIdle(showRootWindow, window) with no retain). 8.6 is no…

      Open ticket aeb2ed51ab · 66. Aqua Window Operations · filed 2025-11-29

      ✅ Since triage (updated 2026-11-11)

      Status now: Closed (Fixed)

      • 🔧 2026-10-24 check-in 996b73f3a3 on macos-root-destroyed-early, in main, 9.0: Fix aeb2ed51ab and 8e32cb7ece: on macOS, destroying "." before the root window was first shown (e.g. [after 0 {destroy .}]) crashed in the showRootWindow()
      • 💬 2026-10-24 serhiy.storchaka: Fixed on branch macos-root-destroyed-early 996b73f3a3, as suggested: window is retained before Tcl_DoWhenIdle() and released in showRootWindow(). The same commit fixes the crash in 8e32cb7ece, which happens once this one is avoided. New test window-2.13 runs the script from there. Tested on Linux and Windows. I have no Mac, so could someone test it on macOS?
      • 💬 2026-10-29 marc_culler: The macos-root-destroyed-early branch stops this crash. I have merged that branch into trunk and will close this ticket.
      • 💬 2026-11-10 nflyernz: ### Independent macOS ARM64 verification of the showRootWindow() fix I encountered a native Tkinter crash while investigating Python 3.15.2 GUI tests in ArduPilot MethodicConfigurator. The problem was reduced to this standalone reproducer, requiring no application code or third-party modules: ```python import tkinter as tk first = tk.Tk() first.destroy() second = tk.Tk() second.update_idletasks() second.destroy(…
      Full verdict

      CANNOT-TEST (macOS). Code check: the bug is still present in main, 9.1 and 9.0 (Tcl_DoWhenIdle(showRootWindow, window) with no retain). 8.6 is not affected: it calls makeKeyAndOrderFront directly and has no idle task.

      • Platform Aqua
      • Tested code reading only (main a695f427fd, tk9.1, tk9.0, tk8.6)
      • Cause macosx/tkMacOSXInit.c:TkpInit passes an NSWindow that is not retained to an idle handler. If after 0 {destroy .} runs first, the window is deallocated before showRootWindow() runs.
      • Prototype aeb2ed51ab.patch (23 changed lines, no test; not compiled, macOS-only file). It retains the window before Tcl_DoWhenIdle and releases it at the end of showRootWindow on every path. It also covers 8e32cb7ece: it skips the work when TkMacOSXGetTkWindow() returns NULL.
      • Suggested ticket comment The window passed to the showRootWindow idle handler was not retained. Proposed fix: retain it in TkpInit and release it in showRootWindow, and skip the work if the Tk window is already gone (see also 8e32cb7ece). Needs a test on macOS with the script from the comment.
      ✅8e32cb7ece2025Aqua: window-2.12 crashpatchCANNOT-TEST (macOS). Code check: still present in main, 9.1 and 9.0. showRootWindow() dereferences the result of TkMacOSXGetTkWindow(root) without …

      Open ticket 8e32cb7ece · 66. Aqua Window Operations · filed 2025-11-28

      ✅ Since triage (updated 2026-11-11)

      Status now: Closed (Fixed)

      • 🔧 2026-10-24 check-in 996b73f3a3 on macos-root-destroyed-early, in main, 9.0: Fix aeb2ed51ab and 8e32cb7ece: on macOS, destroying "." before the root window was first shown (e.g. [after 0 {destroy .}]) crashed in the showRootWindow()
      • 🔧 2026-10-25 check-in 7c3fb7d940 on macos-root-destroyed-early, in main, 9.0: window-2.13: explain why the delay of "destroy ." must be 0, as suggested by chrstphrchvz in 8e32cb7ece.
      • 💬 2026-10-24 serhiy.storchaka: Fixed on branch macos-root-destroyed-early 996b73f3a3, for both this ticket and aeb2ed51ab, as suggested there: the NSWindow is retained until showRootWindow() runs, and the handler does nothing if the Tk window is gone. New test window-2.13 runs the script from this ticket. The test also found a background error on Windows: the console's idle script that erases the "% " prompt ran after the application was de…
      • 💬 2026-10-24 marc_culler: On macOS 27.0.1 with an M3 CPU the script in this ticket runs with no issue with the macos-root-destroyed-early branch.
      • 💬 2026-10-24 chrstphrchvz: I suggest putting a comment in the test above after 0 {destroy .} to say that specifying a delay of 0 is intentional and necessary, in hopes that if the test fails for someone in the future, they won’t try to increase the delay and potentially defeat the purpose of the test. Although I wonder if such a comment makes it good enough to have after 0 {destroy .} in window-2.12 and then not need a separate test.
      • 💬 2026-10-29 marc_culler: I am closing this since it was also fixed by the merge of the macos-root-destroyed-early branch.
      Full verdict

      CANNOT-TEST (macOS). Code check: still present in main, 9.1 and 9.0. showRootWindow() dereferences the result of TkMacOSXGetTkWindow(root) without a NULL check. 8.6 is not affected.

      • Platform Aqua
      • Tested code reading only
      • Cause macosx/tkMacOSXInit.c:showRootWindow. When . is destroyed before the idle handler runs, TkMacOSXGetTkWindow returns NULL, which is then dereferenced. This is the same code path as aeb2ed51ab.
      • Prototype 8e32cb7ece.patch (identical to aeb2ed51ab.patch; fix both together)
      • Suggested ticket comment Fixed together with aeb2ed51ab: showRootWindow now checks for a NULL TkWindow and releases the retained NSWindow. Test window-2.12 should pass on macOS.
      ✅68721906482025Aqua: over-release in CreateCGImageFromDrawableRect()patchCANNOT-TEST (macOS). Code check: still present in main, 9.1 and 9.0 (CGColorSpaceRelease(colorspace) on the result of the Get-rule function CGIma…

      Open ticket 6872190648 · 82. X11 Emulation · filed 2025-11-25

      ✅ Since triage (updated 2026-11-11)

      Status now: Closed (Accepted)

      • 🔧 2026-10-01 check-in 2d659391fc on trunk, in main, 9.0: Merge fix for 6872190648 - Aqua: over-release in CreateCGImageFromDrawableRect()
      • 🔧 2026-10-01 check-in ef133269e4 on core-9-0-branch, in main, 9.0: Backport fix for 6872190648 - Aqua: over-release in CreateCGImageFromDrawableRect() - from Christopher Chavez
      • 🔧 2026-10-08 check-in 2bf909daa9 on bug-df4b603950-alt, in main, 9.0: Cherrypick fix for 6872190648: Aqua: over-release in CreateCGImageFromDrawableRect()
      • 💬 2026-10-01 serhiy.storchaka: Still present in 9.0 and 9.1 (8.6 does not have this code). The fix in branch [bug-6872190648] is correct: CGImageGetColorSpace() follows the Get rule, so its result must not be released. The branch merges cleanly into main. Could someone with macOS test and merge it into core-9-0-branch and main? A probable reproducer (untested, needs a Retina display): [.c postscript] grabs the embedded label with XGetImage(), w…
      • 💬 2026-10-01 marc_culler: Yup. That crashes on my macbook Air M3.
      • 💬 2026-10-01 marc_culler: And removing that call to CGColorSpaceRelease( eliminates the crash. I have merged the fix into 9.0 and 9.1.
      Full verdict

      CANNOT-TEST (macOS). Code check: still present in main, 9.1 and 9.0 (CGColorSpaceRelease(colorspace) on the result of the Get-rule function CGImageGetColorSpace). 8.6 is not affected because it has no downscaling code.

      • Platform Aqua
      • Tested code reading only
      • Cause macosx/tkMacOSXImage.c:CreateCGImageFromDrawableRect releases a colorspace it does not own.
      • Prototype 6872190648.patch (1 line removed, no test)
      • Suggested ticket comment CGImageGetColorSpace() follows the Get rule, so its result must not be released. Proposed fix: remove the CGColorSpaceRelease() call.
      ⬜c36e462daf2025macOS : arrow buttons of ttk::spinbox in dark mode are movingCANNOT-TEST (macOS 26). Code unchanged in main: SpinButtonReBounds() still has bounds->size.height -= 0.5 in dark mode (from e513f55ab9, 2022-10-…

      Open ticket c36e462daf · 66. Aqua Window Operations · filed 2025-10-31

      Full verdict

      CANNOT-TEST (macOS 26). Code unchanged in main: SpinButtonReBounds() still has bounds->size.height -= 0.5 in dark mode (from e513f55ab9, 2022-10-08). The function is not in 8.6.

      • Platform Aqua
      • Tested code reading only
      • Cause macosx/ttkMacOSXTheme.c:SpinButtonReBounds applies a half-pixel height adjustment in dark mode only. Per Marc, this was a workaround for older macOS versions.
      • Prototype none (it needs adjustments that depend on the OS version, and those need testing on several macOS versions)
      • Suggested ticket comment The dark-mode adjustment in SpinButtonReBounds() is still present in main. It should probably be limited to the older macOS versions where it was needed, which needs testing on macOS.
      ⬜b38e430e262025Labels are rendered incorrectly on MacOS when a line starts with \n, and -wraplength is set.CANNOT-TEST (macOS). Not reproduced on X11.

      Open ticket b38e430e26 · 03. [*button] and [label] · filed 2025-09-23

      Full verdict

      CANNOT-TEST (macOS). Not reproduced on X11.

      • Platform Aqua
      • Tested main, 8.6 on X11/Xvfb: reqheight is the same with and without -wraplength 20000 (64/64 and 55/55).
      • Cause probably macosx/tkMacOSXFont.c:Tk_MeasureCharsInContext(). The wrapping branch (maxLength >= 0) trims trailing whitespace ("Trim right whitespace", ~line 1104) and leaves the space before "\n" unconsumed, so the next line starts with " \n" and an empty line is added (chrstphrchvz's analysis). The code is still the same on main.
      • Prototype none (Aqua only, cannot build or verify)
      • Suggested ticket comment Not reproducible on X11 (main, 8.6), so this is Aqua-specific. The likely cause is the whitespace trimming in Tk_MeasureCharsInContext() (tkMacOSXFont.c), which does not consume the space before the newline when wrapping.
      ⬜8369db351b2025Aqua: undefined behavior in GenerateUpdates()patchCANNOT-TEST (macOS). By code reading the problem is still on main: tkMacOSXWindowEvent.c GenerateUpdates() still converts a possibly null (inf) dam…

      Open ticket 8369db351b · 99. Other · filed 2025-08-31

      Full verdict

      CANNOT-TEST (macOS). By code reading the problem is still on main: tkMacOSXWindowEvent.c GenerateUpdates() still converts a possibly null (inf) damageBounds to int.

      • Platform Aqua
      • Tested code inspection only
      • Cause macosx/tkMacOSXWindowEvent.c:GenerateUpdates(). When the window is outside the visible area, CGRectIntersection() returns CGRectNull, whose origin is inf.
      • Prototype 8369db351b.patch (55 lines, mostly re-indentation; branch bug-8369db35 2c134295d9 rebased onto main, sends Expose only if !CGRectIsNull(damageBounds); no test, not compiled)
      • Suggested ticket comment The proposed fix in 2c134295d9 (branch bug-8369db35) still applies to main and has not been merged yet. Needs a macOS build and test run.
      ✅1c9965ca712025possible code issue: return before END_DRAWINGpatchCANNOT-TEST (macOS). By code reading all 3 spots are still present on main: ttkMacOSXTheme.c EntryElementDraw default: return; inside BEGIN/END_D…

      Open ticket 1c9965ca71 · 88. Themed Tk · filed 2025-08-18

      ✅ Since triage (updated 2026-11-11)

      Status now: Closed (Accepted)

      • 🔧 2026-09-30 check-in bc510f94c0 on macos-restore-drawing-context, in main, 9.0: Fix 1c9965ca71: possible code issue: return before END_DRAWING. Restore the drawing context on early returns after it was set up: in EntryElementDraw() for an
      • 🔧 2026-10-01 check-in 89acf2bacf on trunk, in main, 9.0, 8.6: Merge macosx-restore-srawing-context - fixes 1c9965ca71: possible code issue: return before END_DRAWING
      • 🔧 2026-10-01 check-in d0a3c917f0 on core-9-0-branch, in main, 9.0: Backport fix for 1c9965ca71: possible code issue: return before END_DRAWING
      • 🔧 2026-10-01 check-in 26d7a954ab on core-8-6-branch, in main, 9.0, 8.6: Backport fix for 1c9965ca71: possible code issue: return before END_DRAWING
      • 💬 2026-09-30 serhiy.storchaka: Both are bugs: TkMacOSXRestoreDrawingContext() pops the graphics state pushed by TkMacOSXSetupDrawingContext(), releases the clip region and schedules the update of the view, so returning without it leaves an extra saved graphics state, leaks the clip region and misses the update. A search for other returns between a successful TkMacOSXSetupDrawingContext() and TkMacOSXRestoreDrawingContext() found one more: XCopy…
      • 💬 2026-09-30 marc_culler: This looks correct. I guess the CI will run tonight. Assuming the CI run is OK (which it should be - I saw no unexpected test failures on my machine), I think this can be merged.
      • 💬 2026-09-30 marc_culler: I have to correct my statement that I saw no unexpected test errors. There was one which I have not seen before: ==== button-16.1 button keeps -overrelief after click, Bug 1100518 FAILED ==== button-16.1 FAILED I doubt that this patch could affect button tests. Does anyone know anything about that failure?
      • 💬 2026-10-01 serhiy.storchaka: button-16.1 is the test for [1100518], not related to this patch. On Aqua pressing a button does not change its relief (the pressed state is drawn natively), but the test expected "sunken". Fixed in 810f018b3d (main), bc39b6753a (9.0) and a1edb3c002 (8.6).
      • 💬 2026-10-01 marc_culler: Well, the CI run failed because tests font-17.5 and button-16.1 failed. The failure of font-17.1 has been reported by Torsten Berg on the core list and the failure of button-16.1 has apparently been addressed. I do not see font-17.1 failing on my machine, which indicates to me that it is a sporadic failure, probably caused by a race condition. I am sure it is not related to this patch. So I think it is safe to mer…
      • 💬 2026-10-01 marc_culler: I see that the failure of font-17.5 was addressed today in bf9aef00. So I will go ahead and merge this.
      • 💬 2026-10-01 marc_culler: I have merged this patch into core-9-1-branch, core-9-0-branch and core-8-6-branch. So I am closing the ticket. Thanks Serhiy!
      • 💬 2026-10-09 chrstphrchvz: Thanks for getting this fixed. I neglected to comment here about the bug-1c9965ca-c86b and bug-1c9965ca-c90b branches I proposed. However I would still suggest for anyone interested in a ticket to have a look at its timeline view (with "info" in the URL, as opposed to the non-timeline view with "tktview" in the URL) for a better idea of any related activity.
      Full verdict

      CANNOT-TEST (macOS). By code reading all 3 spots are still present on main: ttkMacOSXTheme.c EntryElementDraw default: return; inside BEGIN/END_DRAWING, tkMacOSXFont.c TkpDrawAngledCharsInContext return after SetupDrawingContext when !string, and tkMacOSXImage.c XCopyArea !dc.context return without restore.

      • Platform Aqua
      • Tested code inspection only
      • Cause early returns skip TkMacOSXRestoreDrawingContext() (unbalanced CGContextSaveGState, clipRgn leak)
      • Prototype 1c9965ca71.patch (63 lines). It ports branches bug-1c9965ca-c90b 3e82a60bb2 and bug-1c9965ca-c86b 9bb26ab0c5 to main. For the image hunk it calls TkMacOSXRestoreDrawingContext() before return BadDrawable instead of dropping the return (the branch's version), so nothing draws into a NULL context. Not compiled.
      • Suggested ticket comment The fixes in 3e82a60bb2 and 9bb26ab0c5 are still unmerged. The tkMacOSXImage.c part could keep the return BadDrawable and call TkMacOSXRestoreDrawingContext() before it.
      ⏳c7896924182025tk build on Mac OS X 10.6 fails with "assignment of read-only variable"patchCANNOT-TEST (read core-8-6-16, core-8-6-branch, trunk)

      Open ticket c789692418 · 83. Mac OS X Build · filed 2025-07-24

      ⏳ Since triage (updated 2026-11-11)

      Status now: Pending (Wont Fix)

        Full verdict

        CANNOT-TEST (read core-8-6-16, core-8-6-branch, trunk)

        • Platform Aqua (build, OS X 10.6, gcc 4.2 / llvm-gcc)
        • Tested The failing statement is in -[TKContentView initWithFrame:] in macosx/tkMacOSXWindowEvent.c, at line 981 in 8.6.16 and line 993 on core-8-6-branch: self.layer.contentsGravity = self.layer.contentsAreFlipped ? kCAGravityTopLeft : kCAGravityBottomLeft;. It is unchanged on 8.6 and trunk, and no later check-in mentions 10.6 or gcc 4.2. I could not read the reporter's attachment (patch-tk86-gcc42-macosx-fix.txt, blob eaf0ef3b...): its content is missing from the local repository. According to its description it splits the ternary into if/else, and the reporter says that alone lets 8.6.16 build on 10.6. Marc replied that 10.6 is outside the TIP 715 support range, and resolution is "Wont Fix".
        • Cause gcc 4.2's Objective-C front end turns the property read inside a ?: that is assigned to a property into a read-only temporary (prop.NNN) and then rejects the assignment. This is a compiler bug, not a Tk bug.
        • Prototype c789692418.patch (22 lines, against core-8-6-branch: writes the same assignment as if/else; behaviour unchanged; not compiled, no Mac)
        • Suggested ticket comment 10.6 is not supported (TIP 715), but the reporter's workaround is a behaviour-neutral rewrite of one statement, so it could be applied to 8.6 at no cost before closing. Otherwise close as Wont Fix. Either way the ticket shouldn't stay Pending.
        💬4af5ca19212025XCopyArea is very (very) slow on macOS with version 9CANNOT-TEST (macOS)

        Open ticket 4af5ca1921 · 64. Graphic Contexts · filed 2025-07-23

        💬 Since triage (updated 2026-11-11)

        Status now: Open

        • 💬 2026-10-29 chrstphrchvz: The improvements for [](f215a9fd) were merged, so XCopyArea() will now work again for pixmaps on Tk 9.0+, including the pix.c example. Usage of CGContextSetInterpolationQuality(…, kCGInterpolationNone) which I had mentioned in an earlier comment here has also been added (see [](32f2e19d)). I have not revisited XCopyArea()'s purported performance issues, but in case it needs elaborating, the idea that using `…
        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none (Aqua-only drawing path; the X11 XCopyArea is the real Xlib one)
        • Cause per Marc's analysis in the ticket, on Aqua 9.x XCopyArea/XPutImage to a window ends in one CGContextDrawImage per call (~5-7 ms for 800x400). That is a Quartz limit. It is not a Tk regression that could be fixed; 8.6's drawRect path could not copy from a pixmap at all.
        • Prototype none (no Tk-side fix identified)
        • Suggested ticket comment Per the analysis below, the cost is one CGContextDrawImage per copy, which Tk cannot avoid on macOS. Suggest closing as "Works as designed" unless the reporter gives a case where Tk itself adds overhead.
        ✅307b06cbc22025Gridded widget is not drawn in some cases (Aqua).CANNOT-TEST (macOS); NOT-REPRODUCED on X11 (main, 8.6)

        Open ticket 307b06cbc2 · 49. [grid] · filed 2025-07-18

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Duplicate)

        • 🔧 2026-09-29 check-in 1b1c0ed60e on aqua-unmap-redraw, in main, 9.0: Fix 2ef5dd8036 and 307b06cbc2: Aqua: "pack forget" leaves window on screen. Since 9.0, unmapping a non-toplevel window no longer redrew its toplevel, so the
        • 🔧 2026-10-06 check-in 19dc130ec3 on trunk, in main, 9.0: Fix 2ef5dd8036 and 307b06cbc2: Aqua: "pack forget" leaves window on screen
        • 🔧 2026-10-06 check-in e1e4cbd699 on core-9-0-branch, in main, 9.0: Fix 2ef5dd8036 and 307b06cbc2: Aqua: "pack forget" leaves window on screen
        • 💬 2026-09-29 serhiy.storchaka: As found below, grid remove and other ways of unmapping a window do not redraw the container. The same issue is reported in 2ef5dd8036. Since fe5604ad44, XMapWindow() and XDestroyWindow() schedule the redraw in an idle task, but XUnmapWindow() does not. Proposed fix in branch aqua-unmap-redraw: XUnmapWindow() schedules the same idle redraw. I could not test it, as I have no macOS.
        • 💬 2026-10-06 jan.nijtmans: Dup of 2ef5dd8036 Closing
        Full verdict

        CANNOT-TEST (macOS); NOT-REPRODUCED on X11 (main, 8.6)

        • Platform Aqua
        • Tested main 9.1b1 and 8.6.18 on X11 with no WM. The reporter's script was automated and after each press a screenshot of the lower button was compared with one taken after a forced full re-expose. They were always identical.
        • Cause (from chrstphrchvz's comments): on Aqua 9.x, XUnmapWindow (macosx/tkMacOSXSubwindows.c) no longer makes the parent redraw, and no Expose is generated for the uncovered area. So grid remove/pack forget leaves a stale drawing. X11 gets real Expose events from the server.
        • Prototype none (Aqua-only; the likely fix is TkMacOSXInvalidateWindow/generateExposeEvents on the parent at the end of XUnmapWindow)
        • Suggested ticket comment Not reproducible on X11 (main, 8.6), because the server generates Expose events there. The fix belongs in the Aqua XUnmapWindow: invalidate/expose the uncovered area of the parent.
        ⬜9308ecd92d2024Menus do not render correctly when starting wish via ssh from macOS 15.2CANNOT-TEST (needs XQuartz on macOS 15.2; works with macOS 11.6 on the same Mac)

        Open ticket 9308ecd92d · 82. X11 Emulation · filed 2024-12-17

        Full verdict

        CANNOT-TEST (needs XQuartz on macOS 15.2; works with macOS 11.6 on the same Mac)

        • Platform X11 (remote display on XQuartz)
        • Tested none
        • Cause most likely XQuartz/macOS 15 (the reporter filed XQuartz issue #423). Tk 8.6.12 on Ubuntu is unchanged between the working and the broken setup.
        • Prototype none
        • Suggested ticket comment The Tk side and the X server protocol are unchanged, and this only appears with a newer macOS/XQuartz. It looks like XQuartz issue 423. Please report back if XQuartz fixes it or if it still happens with a current Tk.
        ⬜204a39580b2024Current VC6 compile warnings in core-8-6-branchCANNOT-TEST (no VC6). The ticket is informational and needs no action.

        Open ticket 204a39580b · 85. Win Build · filed 2024-06-18

        Full verdict

        CANNOT-TEST (no VC6). The ticket is informational and needs no action.

        • Platform build (Windows, MS-VC6)
        • Tested read the 8.6 tip source. The code at the warning sites is unchanged:
          • memcpy in tkCanvas.c:4241 (Tk_Uid* const-ness, harmless)
          • the length arguments of textOutProc/getTextExtentPoint32Proc in tkWinFont.c
          • DEFAULT_BORDERWIDTH is defined both in ttkFrame.c (int 2) and in ttkElements.c ("2"), and the VC6 warning suggests several files are compiled together

        The D4002 warnings for -wd4090/-wd4146 only mean that VC6 does not know those options.

        • Cause n/a
        • Prototype none (the warnings are cosmetic)
        • Suggested ticket comment Informational only. The warnings are harmless on VC6. Propose closing.
        ⬜8a601213162024The wm manage command does not work on current macOS versionsCANNOT-TEST (macOS)

        Open ticket 8a60121316 · 66. Aqua Window Operations · filed 2024-05-28

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested X11 main: wm.test -match wm-manage* passes (8 tests, including wm-manage-1.8), so the problem is specific to Aqua.
        • Cause macosx/tkMacOSXWm.c, wm manage/wm forget (NSWindow re-parenting) on macOS 12+
        • Prototype none
        • Suggested ticket comment Nothing to add from the X11 side. The wm-manage tests pass on X11.
        ⬜cb7250f9602024tk_getOpenFile on macOS crashes with double extensionsCANNOT-TEST (macOS), but most likely FIXED since 9.0.2 / 8.6.16 as a duplicate of 080a28104e (crash from a nil UTType): main ce3d0b2059/[e06d95…

        Open ticket cb7250f960 · 37. [tk_get*File] · filed 2024-01-09

        Full verdict

        CANNOT-TEST (macOS), but most likely FIXED since 9.0.2 / 8.6.16 as a duplicate of 080a28104e (crash from a nil UTType): main ce3d0b2059/e06d95ebb3 2025-01-21/24, 8.7 3136e2b81d, 8.6 269ce15793 2025-01-27

        • Platform Aqua
        • Tested code reading only. setAllowedFileTypes() in macosx/tkMacOSXDialog.c now skips a nil [UTType typeWithFilenameExtension:@"tar.gz"] in main, 9.0 and 8.6.
        • Cause typeWithFilenameExtension: returns nil for "tar.gz", and addObject:nil raised an exception. With the fix the filter is silently dropped. The Catalina report (a .tar.gz file shown greyed out) is a separate limitation: macOS matches only the last extension.
        • Prototype none
        • Suggested ticket comment Probably fixed by 080a28104e (nil UTType check in setAllowedFileTypes, in 8.6.16 and 9.0.2). Please retest. Note that a multi-part extension like .tar.gz still cannot work as a filter on macOS, so use .gz instead.
        ✅9814507dcc2023font measure \u4000 crashes on macOS with XQuartzCANNOT-TEST (macOS/XQuartz). Duplicate of 36e379c01b, already worked around: main d2f53df7a8 2024-09-22, 8.7 9c7fbcd94f, 8.6 5c47c3c280 202…

        Open ticket 9814507dcc · 46. Unix Fonts · filed 2023-12-30

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Duplicate)

        • 💬 2026-10-08 serhiy.storchaka: This is the same XQuartz bug as 36e379c01b (XQuartz issue 216). Since d2f53df7a8 (Tk 9.0.0, and 8.6.16), fonts are loaded with BadValue errors from X_OpenFont ignored, so the font is skipped instead of the program exiting.
        Full verdict

        CANNOT-TEST (macOS/XQuartz). Duplicate of 36e379c01b, already worked around: main d2f53df7a8 2024-09-22, 8.7 9c7fbcd94f, 8.6 5c47c3c280 2024-11-13 (XLoadQueryFontNoXError ignores BadValue from X_OpenFont)

        • Platform X11 (XQuartz only)
        • Tested font measure {Courier -10} \u4000 returns 10 on Linux/Xvfb with main and 8.6. The workaround is present in the main, 9.0 and 8.6 sources.
        • Cause XQuartz bug (XQuartz issue #216). XLoadQueryFont of a Monaco iso10646 font raises BadValue.
        • Prototype none
        • Suggested ticket comment Same issue as 36e379c01b (XQuartz #216). The BadValue is now ignored in XLoadQueryFontNoXError since 8.6.16/9.0.0. Please retest; can probably be closed as a duplicate.
        ⬜5b49b1de8d2023Aqua: window manager tests more likely to fail on macOS 14 SonomaCANNOT-TEST (macOS)

        Open ticket 5b49b1de8d · 66. Aqua Window Operations · filed 2023-09-28

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none. Marc Culler adjusted the delays in 2023-12 and reported a clean run on Sonoma.
        • Cause timing races in the Aqua WM tests (unixWm-51.x, wm-stackorder-*, tk-5.x).
        • Prototype none (macOS only)
        • Suggested ticket comment Delays were adjusted in Dec 2023 (clean run reported on Sonoma). Close unless the failures still show up on a current macOS.
        ⬜b7276a7c092023Wish application (git gui) locks-up systemCANNOT-TEST (a full system lockup with no VT switch is a kernel/i915/Xwayland problem that no X client can cause by itself; no repro script; hardwa…

        Open ticket b7276a7c09 · 75. wish · filed 2023-07-18

        Full verdict

        CANNOT-TEST (a full system lockup with no VT switch is a kernel/i915/Xwayland problem that no X client can cause by itself; no repro script; hardware-specific)

        • Platform X11 (Xwayland on Linux, Intel i915)
        • Tested nothing (no repro; reporter never followed up after 2023-07)
        • Cause unknown, most likely a GPU driver or kernel bug triggered by ordinary X11 traffic
        • Prototype none
        • Suggested ticket comment A user-space X client cannot lock up the whole system, so this points to the kernel, the i915 driver or Xwayland. Please report it there, with a reproducer if you can get one. Closing as not a Tk bug unless a Tk-specific trigger turns up.
        💬c0662eca6e2023On aqua, the after callbacks are delayed until releasing the mouse buttonCANNOT-TEST (macOS)

        Open ticket c0662eca6e · 66. Aqua Window Operations · filed 2023-06-04

        💬 Since triage (updated 2026-11-11)

        Status now: Open

        • 💬 2026-10-08 serhiy.storchaka: This has the same cause as 3ceb6b0f65: during a live resize AppKit runs its own event loop, and only the redraw is done there. With the drawing to a bitmap in 9.0 (1dfe4fe90a) flicker may no longer be a reason to suspend other events. Marc, could TKBackgroundLoop, which keeps Tk events running while a menu is open, also be run during a live resize (between viewWillStartLiveResize and viewDidEndLiveResize), as …
        • 💬 2026-10-10 marc_culler: I think the AppKit handles open menus and window resizing in very similar ways. While a menu is open Appkit runs a runloop with mode NSEventTrackingMode. Gemini claims that the same is true during a liveResize (but I haven't checked that). In both cases there are starting and ending notifications, and we subscribe to both. So I would say that it is very likely possible to run the backgroundLoop thread during a liv…
        • 💬 2026-10-10 marc_culler: I think the guard in deferredUpdateLayer is now correct. I added a boolean property windowIsDead to the TKContentView class and I set that to YES in tkWmDeadWindow. I check the property in deferredUpdateLayer. I think this fix is now ready to merge, but I am going to ask Nicolas Bas to test against his app. His report of artifacts when the main window of his app is withdrawn, configured, and deiconified was the or…
        • 💬 2026-10-10 marc_culler: Please ignore my last comment which was intended for a different ticket but got submitted here by mistake.
        • 💬 2026-10-10 chrstphrchvz: _runBackgroundLoop is only ever called from the main thread, not a background thread. All that the background thread does is repeatedly call performSelectorOnMainThread:… to tell the main thread to eventually call _runBackgroundLoop. If there are still reasons not to perform certain AppKit tasks or handle Tk event callbacks from _runBackgroundLoop, being run from a background thread is not one of them.
        • 💬 2026-10-10 marc_culler: OK. Good. How do the configure events which are handled immediately in setFrameSize trigger the procs which are bound to ? And why don't the procs run when the event is handled? Why does Tcl wait until the Tcl event loop starts up again to run them? Where are those procs stored while the Tcl event loop is stalled due to the runloop mode being NSEventTrackingRunloopMode?
        • 💬 2026-10-11 marc_culler: Is it possible that the messages are written to stdout and buffered because stdout cannot be flushed until the Tcl event loop resumes, so all of the buffered text gets printed when that happens?
        • 💬 2026-10-11 chrstphrchvz: If I add puts to the bind script: `` set count 0 bind . { puts "bind $count" after 100 [list puts " [incr count]"] } ` and then resize the window, I observe the bind script outputting a line immediately each time it is run. The output from the after 100 …` script does not appear until the live resize is over, because timer events are not handled until then. So I do not think output buffering is relevant here…
        • 💬 2026-10-12 chrstphrchvz: Typo in my previous comment: I meant _runBackgroundLoop, not _runLoopMode. I have pushed a branch aqua-liveresize-runbackgroundloop-experiment resembling what I tried a few years ago. There is a related issue to be aware of, and which my branch does not affect. A comment in NSWindow.h (still present in the macOS 26.5 SDK) says: > NSWindowWillStartLiveResizeNotification is sent when the user starts a live resiz…
        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none. The reporter says X11 and Windows behave correctly. macosx/tkMacOSXNotify.c still says "We must not steal any events during LiveResize", and the live-resize code is essentially unchanged (culler's cgimage_drawing_tk9 work from 2024 changed only redrawing).
        • Cause AppKit's live-resize tracking loop blocks, and Tk services only redraws during it (see chrstphrchvz's comment; the temporary runloop used for posted menus could be reused).
        • Prototype none (macOS only, design-level change)
        • Suggested ticket comment none; leave it for the Aqua maintainers.
        ⬜132dd3d3502023Aqua: TIP 655 causes overrideredirect toplevel moved offscreen to be blankCANNOT-TEST (macOS). Probably still present: the TkMacOSXMakeRealWindowExist code ([window setFrame:geometry display:YES] plus the DocModal style…

        Open ticket 132dd3d350 · 66. Aqua Window Operations · filed 2023-05-05

        Full verdict

        CANNOT-TEST (macOS). Probably still present: the TkMacOSXMakeRealWindowExist code ([window setFrame:geometry display:YES] plus the DocModal style mask, tkMacOSXWm.c:~6744/6866) is unchanged in main, and fvogel confirmed the bug on Ventura.

        • Platform Aqua
        • Tested none. IDLE has already removed its +10000+10000 trick (cpython PR 104591).
        • Cause per the reporter, setFrame:display: with an offscreen frame, then adding NSWindowStyleMaskDocModalWindow (TIP 655), means drawRect: is never called.
        • Prototype none (macOS only)
        • Suggested ticket comment none.
        🟢1ab7a5f2992023Aqua: use-after-free in handleReopenApplicationEvent:withReplyEvent:CANNOT-TEST (macOS). A fix exists but is not merged: branch bug-1ab7a5f2995 (culler, f19560a2c1 2026-06-25 and 1eb8885bc7 2026-06-28, based o…

        Open ticket 1ab7a5f299 · 66. Aqua Window Operations · filed 2023-04-25

        🟢 Since triage (updated 2026-11-11)

        Status now: Open

        • 🔧 2026-11-05 check-in 33897dd878 on trunk, in main, 9.0: Fix 1ab7a5f299: Aqua use-after-free
        • 🔧 2026-11-05 check-in 6685f46c8c on core-9-0-branch, in main, 9.0: Fix 1ab7a5f299: Aqua use-after-free
        • 💬 2026-10-08 chrstphrchvz: I was not familiar with Tcl_CallWhenDeleted(). That actually seems better than preserving the interpreter. I don’t consider myself a good person to ask regarding what the behavior should be after deleting _eventInterp; I am currently not that interested in trying to figure out a satisfactory alternative. Would assigning NULL here be a good idea? ``` --- macosx/tkMacOSXInit.c +++ macosx/tkMacOSXInit.c @@ -66,6 …
        Full verdict

        CANNOT-TEST (macOS). A fix exists but is not merged: branch bug-1ab7a5f2995 (culler, f19560a2c1 2026-06-25 and 1eb8885bc7 2026-06-28, based on core-9-0-branch). It registers a Tcl_CallWhenDeleted handler that sets _eventInterp to NULL. Main still stores _eventInterp with no delete callback (tkMacOSXInit.c _setup:).

        • Platform Aqua
        • Tested none (Aqua only). The fix is plausible: every _eventInterp use in tkMacOSXHLEvents.c and tkMacOSXMenus.c checks for NULL.
        • Cause NSApp keeps a raw pointer to the first Tk interpreter. After interp delete, the Apple-event and menu handlers use the freed interpreter.
        • Prototype none (the existing branch covers it)
        • Suggested ticket comment The fix is in branch bug-1ab7a5f2995. It needs review and testing on macOS (the tclsh repros in the 2025-11-04 comments), then merging to 9.0/main and backporting to 8.6.
        ⬜a64a4b331e2022Scale does not appear until window clicked inCANNOT-TEST (macOS)

        Open ticket a64a4b331e · 66. Aqua Window Operations · filed 2022-09-20

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested history only. Fix branch bug-a64a4b331e (single check-in 151f128071, 2022-09-21, XMapWindow generates Expose once for the first mapped toplevel) was never merged; neither 8.6 tip nor main contain it. Main switched to cgimage drawing (9.0), which may have changed the behaviour; related f16cdb6d04 is still open.
        • Cause macosx/tkMacOSXSubwindows.c:XMapWindow - no expose is generated for the initial map of "." until the first user event (per Marc's analysis).
        • Prototype none (existing branch bug-a64a4b331e).
        • Suggested ticket comment Branch bug-a64a4b331e was never merged. Can someone on macOS check whether this still happens with 9.0 (cgimage drawing) and 8.6.17+?
        💬fb13231f6e2022Messagebox can crash app on macCANNOT-TEST (macOS); probably fixed in 9.0

        Open ticket fb13231f6e · 38. [tk_messageBox] · filed 2022-09-13

        💬 Since triage (updated 2026-11-11)

        Status now: Open

        • 💬 2026-10-08 serhiy.storchaka: This is the same kind of problem as 33b3af1cea: a modal dialog opened from a callback that runs inside an AppKit operation, here drawRect:, during which the dialog's event loop cannot run. Since 1dfe4fe90a (Tk 9.0), Tk on macOS draws into its own bitmap instead of in drawRect:, so this should no longer crash in 9.0 and later. In 8.6, open the dialog from after idle. Marc, can you confirm with 9.0?
        • 💬 2026-10-08 chrstphrchvz: This ticket is for Tk crashing rather than becoming unresponsive. I would say it is more related to e.g. 1b286c99ae and c03b8eb71f than it is to 33b3af1cea. The move away from using drawRect: does likely impact this ticket, however.
        • 💬 2026-10-08 marc_culler: That is a pretty bizarre script. Who grids a vertical scrollbar for a Text widget below the widget? Who wants to display an Alert message before the root window appears? Who generates error dialogs in a scrollcommand? Nonetheless, I ran the python code block in Python 3.13.0 on Intel macOS Sequoia. Python 3.13 reports: >>> tkinter.Tk().eval('info patchlevel') '8.6.14' There is no crash. There is a TclError excepti…
        • 💬 2026-10-08 marc_culler: I installed python 3.14.8 on the same computer. It ships with Tk 9.0.4. When I run the first python script with Python 3.14.8 I get the same TclError traceback (not duplicated), but the dialog opens correctly. If I add the call to root.update_idletasks() the exception does not occur.
        • 💬 2026-10-08 serhiy.storchaka: Since aa35622f95 (fix for [991849], merged into main, core-9-0-branch and core-8-6-branch on 2026-10-05, after 9.0.4), the text, canvas and ttk widgets no longer call their scroll commands before the window is mapped, as the listbox already did. In this script the -yscrollcommand is then only called after the window is shown, which is what the update_idletasks workaround does. Marc, could you retry with the curr…
        Full verdict

        CANNOT-TEST (macOS); probably fixed in 9.0

        • Platform Aqua
        • Tested code reading. In 8.6 the text widget's -yscrollcommand runs from DisplayText called from -[TKContentView drawRect:]→generateExposeEvents, and [NSAlert runModal] throws inside a CA transaction. Since 9.0 (cgimage drawing, merged 1dfe4fe90a 2024-07-08) TKContentView no longer draws in drawRect:, so this path should be gone; 8.6 still has it.
        • Cause macosx/tkMacOSXWindowEvent.c drawRect: → Tcl idle processing → script → modal dialog in the drawing transaction.
        • Prototype none (can't test; a guard would be deferring the script or refusing runModal while [NSApp isDrawing]).
        • Suggested ticket comment The crash happens because the -yscrollcommand script runs from drawRect:. Tk 9.0 doesn't draw from drawRect: any more; can someone on macOS check whether 9.0 still crashes? Workaround for 8.6: after idle tk_messageBox ....
        ⬜e0c7091f2b2022Hang handling window resize in display update in callback from another threadCANNOT-TEST (macOS; needs live resize with the mouse). Reporter says Linux is fine.

        Open ticket e0c7091f2b · 66. Aqua Window Operations · filed 2022-09-08

        Full verdict

        CANNOT-TEST (macOS; needs live resize with the mouse). Reporter says Linux is fine.

        • Platform Aqua
        • Tested none.
        • Cause (per chrstphrchvz) stuck in while (Tcl_ServiceEvent(TCL_WINDOW_EVENTS|TCL_DONT_WAIT)) in generateExposeEvents: during live resize, because Thread's ThreadEventProc ignores the flags and evaluates the update script sent by the worker.
        • Prototype none.
        • Suggested ticket comment Needs a macOS retest with 9.0, where live resize drawing was reworked (cgimage drawing).
        ⬜aad0231f072022On macOS, DisplayText is called twice (inside and outside of drawRect)CANNOT-TEST (macOS). This is the design of the Aqua drawing model (drawing only inside drawRect), not something that exists on X11.

        Open ticket aad0231f07 · 66. Aqua Window Operations · filed 2022-06-07

        Full verdict

        CANNOT-TEST (macOS). This is the design of the Aqua drawing model (drawing only inside drawRect), not something that exists on X11.

        • Platform Aqua
        • Tested none
        • Cause macOS port: any display proc called outside drawRect only marks the area dirty. drawRect then redraws the whole widget without logging (see Marc's comment and 3e9e82bcfe).
        • Prototype none (Aqua-only architectural issue)
        • Suggested ticket comment none (needs a macOS developer).
        ⬜adb63528462022space-newline shows as double newline in messageCANNOT-TEST (macOS). X11 is not affected.

        Open ticket adb6352846 · 14. [message] · filed 2022-05-31

        Full verdict

        CANNOT-TEST (macOS). X11 is not affected.

        • Platform Aqua
        • Tested X11 main and 8.6: message with "foo \nbar" has the same reqheight as "foo\nbar" (2 lines).
        • Cause macOS TkpMeasureCharsInContext (macosx/tkMacOSXFont.c) handling of trailing whitespace before a newline. Branch bug-adb6352846 (1450173ffa, 790b818270, test message-4.3 in bd43a38e6c, June 2022) was never merged. Per fvogel the last attempt still did not fix it.
        • Prototype none (Aqua-only)
        • Suggested ticket comment Not reproducible on X11 (main, 8.6). The bug-adb6352846 branch is still unmerged, and its test message-4.3 is the one to check on macOS.
        💬3ceb6b0f652022Aqua: timer events stop working during tk_messageBoxCANNOT-TEST (macOS). X11 is not affected.

        Open ticket 3ceb6b0f65 · 38. [tk_messageBox] · filed 2022-05-01

        💬 Since triage (updated 2026-11-11)

        Status now: Open

        • 💬 2026-10-08 serhiy.storchaka: This is still the case: Tcl's macOS notifier no longer services Tcl events in nested run loops since the 8.6.11 rework, so nothing runs during NSModalPanelRunLoopMode. For menus, Tk already keeps Tk events running during NSEventTrackingRunLoopMode with TKBackgroundLoop. Marc, would it make sense to use the same for NSModalPanelRunLoopMode while a dialog is open? That would make timers work during dialogs as on oth…
        Full verdict

        CANNOT-TEST (macOS). X11 is not affected.

        • Platform Aqua
        • Tested X11 main and 8.6: after 100 timer keeps firing during tk_messageBox (10 ticks/s).
        • Cause per chrstphrchvz, probably the Tcl 8.6.11 macOS_hangs change that removed Tcl_ServiceAll() from UpdateWaitingListAndServiceEvents (Tcl side).
        • Prototype none (Aqua-only)
        • Suggested ticket comment Timers keep running during tk_messageBox on X11 (main, 8.6), so this is Aqua-specific.
        ⬜c0f3263ed12022Spurious drag after dismissing menuCANNOT-TEST (macOS). NOT-REPRODUCED on X11 and Windows.

        Open ticket c0f3263ed1 · 01. Bindings · filed 2022-04-30

        Full verdict

        CANNOT-TEST (macOS). NOT-REPRODUCED on X11 and Windows.

        • Platform Aqua (reported on macOS)
        • Tested Xvfb with real XTEST input (small XTest helper, Xvfb only), no WM and twm; main, 8.6 and 8.4. After tk_popup, click in the toplevel while moving to dismiss: 0 B1-Motion events during the click, and 0 after release while moving without a button.
        • Tested on Windows (2026-10-16): Windows 11 VM, real input with SendInput (work-win/wmseq.c), 9.1 with tk_popup and with $m post, 8.6.18 with tk_popup; a screenshot confirmed the native menu was posted. Press in the toplevel, drag, release, move on: the dismissing press is delivered to the window (10 B1-Motion during the drag, state 256), but 0 B1-Motion after the release and 10 plain Motion events.
        • Cause unknown (Aqua event handling after the menu's modal tracking loop)
        • Prototype none
        • Suggested ticket comment Not reproducible on X11 (main, 8.6) and Windows (9.1, 8.6): no B1-Motion events after dismissing the menu by clicking in the window. Looks Aqua-specific.
        🟢f16cdb6d042022Aqua: widgets not mapped immediatelyCANNOT-TEST (macOS)

        Open ticket f16cdb6d04 · 66. Aqua Window Operations · filed 2022-04-09

        🟢 Since triage (updated 2026-11-11)

        Status now: Open

        • 🔧 2026-10-08 check-in 1269dd2ba9 on bug-40dc19ee9e, in main, 9.0: Fix 40dc19ee9e: Canvas does not immediately display and f16cdb6d04: Aqua widgets not mapped immediately
        • 💬 2026-10-08 serhiy.storchaka: This looks like the same problem as 40dc19ee9e: on macOS the widgets created by a script are not displayed, and their display procedures (which call -yscrollcommand) are not run, until an event or update idletasks arrives. Since this was reported for 8.6.12, it is older than the change bisected there. A fix for that ticket should be checked with this script too.
        • 💬 2026-10-08 chrstphrchvz: While this ticket and a64a4b331e (and maybe others) do have similar symptoms and workarounds as 40dc19ee9e, they were observed before Tk Aqua moved away from drawRect:. If the issues in the older tickets still seem present, the exact cause is probably now different. For the example in this particular ticket, what I notice with recent Tk 9.1 is that, even though the listbox is not visible right away, the error …
        • 💬 2026-10-09 chrstphrchvz: To further argue that the issue seen under 8.6 is distinct from the one seen in recent 9.0/9.1, consider an amended example: . configure -background red pack [listbox .l -yscrollcommand bogus] .l insert 0 foo With recent 8.6, the red toplevel background is visible before forcing the listbox to appear (e.g. by clicking or resizing); whereas with recent 9.0/9.1, neither the listbox nor the red toplevel background ar…
        • 💬 2026-10-09 marc_culler: Here is another argument. As far as I can tell both your original script and the amended version work fine in the bug-40dc19ee9e branch. The fix in that branch changes nothing except for the updateLayer method in 9.1. Since 8.6 does not even have an updateLayer method at all, the 8.6 issue could hardly be the same as the 9.1 issue. When I run the scripts in bug-40dc19ee9e the toplevel never turns red but the listb…
        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested X11 main, 8.6 — bgerror "invalid command name bogus" raised immediately (no issue on X11; fvogel confirmed Windows OK). No later check-in in the timeline mentions this.
        • Cause unknown (Aqua-specific: widget window not mapped/displayed until an event arrives, so the listbox never runs its display/scroll-update code)
        • Prototype none
        • Suggested ticket comment Not reproducible on X11 (main, 8.6); Aqua-specific.
        ⬜71a2cadbca2022macOS Aqua: behaviour of Drag operationsCANNOT-TEST (macOS; trackpad tap-to-drop not delivering <ButtonRelease-1>, needs TkDnD + trackpad)

        Open ticket 71a2cadbca · 66. Aqua Window Operations · filed 2022-01-06

        Full verdict

        CANNOT-TEST (macOS; trackpad tap-to-drop not delivering <ButtonRelease-1>, needs TkDnD + trackpad)

        • Platform Aqua
        • Tested n/a
        • Cause unknown (reporter narrowed it to trackpad tap not generating ButtonRelease-1)
        • Prototype none
        • Suggested ticket comment n/a (macOS trackpad-specific).
        💬25757f142e2021Tk not found in threads in 8.7 single file build binariesCANNOT-TEST (the VM has no Thread package and no zipfs single-file wish; building both is too costly for this run)

        Open ticket 25757f142e · 85. Win Build · filed 2021-12-22

        💬 Since triage (updated 2026-11-11)

        Status now: Open

        • 💬 2026-10-20 serhiy.storchaka: Still reproducible on Windows with the current main (9.1), in wish91.exe built with win/makefile.vc and not installed, with Thread 3.0.7: package require Thread set t [thread::create] thread::send $t {package require tk} ;# can't find package tk thread::send $t {load .../tcl9tk91.dll Tk} ;# works Tk itself works in the new thread; it finds its library in the zip archive of tcl9tk91.dll (//zipfs:/lib/tk/tk_library)…
        Full verdict

        CANNOT-TEST (the VM has no Thread package and no zipfs single-file wish; building both is too costly for this run)

        • Platform Windows (build)
        • Tested Linux zipfs builds (Tk lib mounted from //zipfs:/lib/tk), main 9.1b1 and 9.0.5, with Thread 3.0.7: package require Tk in a thread works (tk_library=//zipfs:/lib/tk/tk_library). The Windows single-file (zip appended to the exe) CI binaries could not be tested. Side note: with 9.0.5 wish the process crashed with SIGBUS at exit after loading Tk in a second thread (main was fine). Not investigated.
        • Cause unknown. Probably the per-thread tk_library/zipfs mount lookup in the single-file exe case (TkpGetAppName/TclZipfs_AppHook only run in the main thread).
        • Prototype none (Windows-only). Repro script: win/25757f142e.tcl (needs a Thread package; prints a VERDICT line)
        • Suggested ticket comment On Linux zipfs builds (9.0, 9.1b1), package require Tk in a Thread-created thread works. It still needs to be checked with the Windows single-file binaries.
        Windows VM, run 3
        • Platform Windows (build)
        • Verdict CANNOT-TEST (the VM has no Thread package and no zipfs single-file wish; building both is too costly for this run)
        • Tested none (tcl91\pkgs is empty, and there is no thread*.dll)
        • Cause unknown
        • Prototype none
        • Suggested ticket comment (none yet; Linux zipfs builds work, see result-14)
        ✅a4c72a04e32021macOS Aqua : little error with menu-buttons demopatchCANNOT-TEST (macOS). The script-level cause was confirmed on X11 by emulating a modal post that destroys the toplevel.

        Open ticket a4c72a04e3 · (unused) · filed 2021-12-17

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Fixed)

        • 🔧 2026-10-08 check-in cda795ec1c on menubutton-post-destroyed, in main, 9.0: Fix a4c72a04e3: errors when a menu is destroyed while it is posted. The menu can be destroyed during "$menu post", by the -postcommand or, on macOS, while the
        • 💬 2026-10-08 serhiy.storchaka: Whatever Cmd+W should do, the error is a bug: the menu can be destroyed while it is posted, during $menu post: on macOS while the native menu is shown, and on all platforms by the -postcommand. tk::PostOverPoint, tk::MbPost and tk_popup then used the destroyed menu. On X11 the same errors can be triggered with a -postcommand which destroys the toplevel. Proposed fix in branch menubutton-post-destroyed, with tests …
        • 💬 2026-10-15 jan.nijtmans: > Tested on X11 and Windows; not on macOS On macOS: ---- menu-46.1 start /bin/sh: line 1: 15406 Segmentation fault: 11 ./tktest /Users/runner/work/tk/tk/tk/unix/../tests/all.tcl -file menu.test -verbose t make: *** [test-classic] Error 139 Since this fix is Tcl-only, it shows that there's an additional problem on macOS. I don't see anything wrong with the branch. Therefore I temporarily added a "notAqua" constrain…
        • 💬 2026-10-16 marc_culler: I just tested this with the tip_765 branch on M4 macOS 26. This problem seems to be fixed in that branch. For every menu in the menu-buttons demo, typing Cmd+W while the menu is displayed causes the window to close with no error. I don't know which part of the TIP 765 implementation fixed this problem, but it seems like another good reason to approve and merge that TIP.
        • 💬 2026-10-16 marc_culler: At least the error is gone. I still don't know what the correct behavior of Cmd+W should be when a menu is open. I guess it depends on whether you consider the menu to be the frontmost window or just a part of the frontmost window. However, Tk never receives any key events while a menu is open because AppKit is running a runloop with runloop mode NSEventTrackingRunLoopMode while the menu is open, and that runloop …
        • 💬 2026-10-17 nemethi: Merged by Jan into main and core-9-0-branch by commits 6037383f and 0168135c; backported to core-8-6-branch by commit 8cbb336f.
        • 💬 2026-10-22 jan.nijtmans: > I don't know which part of the TIP 765 implementation fixed this problem, > but it seems like another good reason to approve and merge that TIP. The problem with menu-46.1 is fixed in 4bbb759d follow-up commit
        Full verdict

        CANNOT-TEST (macOS). The script-level cause was confirmed on X11 by emulating a modal post that destroys the toplevel.

        • Platform Aqua (the script-level cause is generic)
        • Tested main X11 with the post subcommand wrapped so it destroys the toplevel (emulates Cmd+W during Aqua's modal menu tracking). The unpatched library raises "invalid command name", the patched one does not.
        • Cause library/menu.tcl: tk::PostOverPoint. On aqua (and win32) $menu post is modal. If the window is destroyed during tracking, the following $menu type/entrycget $entry runs on a dead menu.
        • Prototype a4c72a04e3.patch (8 changed lines, adds a winfo exists $menu check after post in both PostOverPoint variants; no test; menu/menubut/menuDraw tests pass)
        • Suggested ticket comment The error comes from tk::PostOverPoint, which uses the menu after the modal $menu post returns, when the menu may already have been destroyed. Checking winfo exists $menu there fixes it. This does not change what Cmd+W does.
        ⬜6a33585a352021Legacy and Revised [text]: textDisp-6.5, -6.6, -7.[1-8] fail on aquaCANNOT-TEST (macOS)

        Open ticket 6a33585a35 · 18. [text] · filed 2021-07-24

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested n/a. In main and 8.6 these tests are constrained aquaKnownBug. Per fvogel's analysis, the cause is the different damage region of TkScrollWindow on macOS, plus lines redrawn inside drawRect not being logged.
        • Cause macosx TkScrollWindow damage region (source minus destination rectangle) and OK_TO_LOG suppressed inside drawRect.
        • Prototype none
        • Suggested ticket comment The tests are marked aquaKnownBug in core-8-6-branch and main. What remains is a test/logging artefact on Aqua, so this could be closed as "Works as designed" unless someone wants to make the logging match.
        ⬜8d42fe7ccc2021Legacy [text]: (textDisp-2.26 and 24.24) tag -tabs does not work properly on aquaCANNOT-TEST (macOS)

        Open ticket 8d42fe7ccc · 18. [text] · filed 2021-07-24

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested n/a. textDisp-2.26 and textDisp-24.24 are constrained notAqua in both main and 8.6.
        • Cause unknown (right-justified tab handling in the Aqua text layout)
        • Prototype none
        • Suggested ticket comment Still open on Aqua. Both tests are constrained notAqua in 8.6 and main. Needs checking on macOS.
        ⬜e1e881b9202021Aqua Ttk entry: only foreground color customizable under dark modeCANNOT-TEST (macOS)

        Open ticket e1e881b920 · 88. Themed Tk · filed 2021-03-01

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested code reading only. In macosx/ttkMacOSXTheme.c:EntryElementDraw on macOS > 10.8, DrawEntry uses the system text background and ignores -fieldbackground. The -fieldbackground/-background handling exists only in the legacy HITheme branch. The foreground is still customizable.
        • Cause see above. The fix is a policy choice: either ignore custom -foreground on Aqua (as Marc suggested) or honour -fieldbackground in DrawEntry.
        • Prototype none (macOS only)
        • Suggested ticket comment (for a macOS tester) EntryElementDraw in ttkMacOSXTheme.c ignores -fieldbackground on 10.9+. We need a decision: honour -fieldbackground in DrawEntry, or also ignore -foreground.
        ✅106af72ae02021windows: wm attributes .window -fullscreen puts screen on a different monitorCANNOT-TEST for two monitors: the VM has one monitor, and changing the VirtualBox monitor count needs the VM to be powered off. The prototype compi…

        Open ticket 106af72ae0 · (unused) · filed 2021-02-17

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Duplicate)

        • 🔧 2026-10-03 check-in 94e8295d08 on win-fullscreen-monitor, in main, 9.0: Fix 4face822b0: wm attributes -fullscreen w/dual monitor opens to fullscreen in the wrong display. UpdateWrapper always used the primary screen; use the monit
        • 💬 2026-10-03 serhiy.storchaka: Duplicate of 4face822b0. Proposed fix in branch win-fullscreen-monitor.
        Full verdict

        CANNOT-TEST for two monitors: the VM has one monitor, and changing the VirtualBox monitor count needs the VM to be powered off. The prototype compiles and does not change single-monitor behaviour.

        • Platform Windows
        • Tested none on Linux (X11 fullscreen is done by the WM via _NET_WM_STATE_FULLSCREEN). Repro script: win/106af72ae0.tcl (prints VERDICT, and CANNOT-TEST when only one monitor is detected).
        • Cause win/tkWinWm.c:UpdateWrapper, WM_FULLSCREEN branch, hardcodes x = 0; y = 0; width/height = WidthOfScreen/HeightOfScreen(...), which is the primary monitor. Fix: get the monitor from the old wrapper (MonitorFromWindow(oldWrapper, MONITOR_DEFAULTTONEAREST), or MonitorFromPoint at winPtr->changes.x/y if it was never mapped), then use GetMonitorInfoW(...).rcMonitor for x/y/width/height. About 15 lines.
        • Prototype none (Windows only; fix described above)
        • Suggested ticket comment On Windows, fullscreen mode always uses the origin and size of the primary screen (UpdateWrapper in tkWinWm.c). It should use the rcMonitor of the monitor the window is on (MonitorFromWindow/GetMonitorInfo).
        Windows VM, run 3
        • Platform Windows
        • Verdict CANNOT-TEST for two monitors: the VM has one monitor, and changing the VirtualBox monitor count needs the VM to be powered off. The prototype compiles and does not change single-monitor behaviour.
        • Tested 4face822b0.patch built in tk-wt91 (no warnings). work-win/fs.tcl on ref91 and on the prototype: normal→fullscreen→normal gives 1023x768+0+0 then back to 300x200+120+90; zoomed→fullscreen→normal gives 1023x768+0+0 then back to zoomed. Both builds give the same results; the only difference is the zoomed position, which comes from the 557441302b patch in the same build. wm.test, winWm.test, winDialog.test, menu.test and winMenu.test: 1096 tests, 0 failed.
        • Cause win/tkWinWm.c UpdateWrapper, WM_FULLSCREEN branch (x=y=0, WidthOfScreen/HeightOfScreen)
        • Prototype 4face822b0.patch (unchanged, 27 lines; compiles cleanly with MSVC)
        • Suggested ticket comment A fix using MonitorFromRect/GetMonitorInfo is ready. It is verified on a single monitor, where behaviour does not change, but still needs someone with two monitors to test it.
        ⬜94bdc4cc672021macos: virtual screen information is not reported.CANNOT-TEST (macOS; mostly an RFE for a multi-monitor API)

        Open ticket 94bdc4cc67 · (unused) · filed 2021-02-16

        Full verdict

        CANNOT-TEST (macOS; mostly an RFE for a multi-monitor API)

        • Platform Aqua
        • Tested code reading only. macosx/tkMacOSXWm.c:Tk_GetVRootGeometry returns wmPtr->vRoot*, filled in by UpdateVRootGeometry.
        • Cause not investigated further. The request for a list of screens and their geometry is a TIP-level RFE.
        • Prototype none
        • Suggested ticket comment The multi-monitor query part is an RFE (it needs a TIP). A macOS tester should check whether winfo vroot* reports the union of screens.
        ⬜c2fc80fe532021macos: withdrawing a fullscreen mode window leaves the screen blackCANNOT-TEST (macOS)

        Open ticket c2fc80fe53 · (unused) · filed 2021-02-16

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none
        • Cause likely wm withdraw orders out the NSWindow while it is still in its own fullscreen Space. The fix would be to leave fullscreen (toggleFullScreen) and withdraw only after windowDidExitFullScreen, or to refuse the withdraw.
        • Prototype none
        • Suggested ticket comment This needs a macOS tester to check whether it still happens with current 8.6/9.0.
        ⬜7e33bcfd9a2021mac os : error messages during -fullscreenCANNOT-TEST (macOS)

        Open ticket 7e33bcfd9a · (unused) · filed 2021-02-16

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none (the "not in fullscreen state" message comes from AppKit toggling native fullscreen; Ctrl-Cmd-F is a macOS menu shortcut)
        • Cause unknown; probably a race between Tk's -fullscreen toggling (tkMacOSXWm.c) and AppKit's native fullscreen transition
        • Prototype none (macOS only)
        • Suggested ticket comment Needs retesting on current macOS with 9.0/8.6 tip; not testable on Linux/Windows.
        ⬜fa1a4cfa2a2020Aqua: revisiting idle drawing issuesCANNOT-TEST (macOS)

        Open ticket fa1a4cfa2a · 02. Appearance · filed 2020-12-11

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none
        • Cause macosx/tkMacOSXDraw.c/tkMacOSXWindowEvent.c: drawRect: is sometimes called outside TkMacOSXDrawAllViews(), which reenters idle processing. This is an investigation ticket with no concrete misbehaviour.
        • Prototype none
        • Suggested ticket comment Needs a macOS developer. The Aqua drawing code has changed a lot since 2020 (the drawing model was rewritten for 8.6.11+/9.0), so it should be rechecked against 9.0 or closed.
        ⬜9ac1b02e242020macOS: duplicate -lpthread -framework CoreFoundation flagsCANNOT-TEST (macOS build)

        Open ticket 9ac1b02e24 · 83. Mac OS X Build · filed 2020-08-08

        Full verdict

        CANNOT-TEST (macOS build)

        • Platform build
        • Tested main unix/configure on Linux. LIBS/WISH_LIBS contain -lpthread only once, so Linux is not affected. The duplicates probably come from the Darwin branch of SC_CONFIG_CFLAGS / CoreFoundation check in unix/tcl.m4 (LIBS="$LIBS -framework CoreFoundation") plus the flags inherited from Tcl.
        • Cause unknown without a macOS build. Cosmetic only: the linker ignores the duplicates.
        • Prototype none
        • Suggested ticket comment Harmless and not seen on Linux. Needs someone with a macOS build to check where the second copy is added (tcl.m4 Darwin/CoreFoundation section).
        ✅bb5b343fe12020freebsd: test failuresCANNOT-TEST (FreeBSD). NOT-REPRODUCED on Linux.

        Open ticket bb5b343fe1 · 86. Test Tools · filed 2020-07-23

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Fixed)

        • 🔧 2026-09-27 check-in 5af8f676bc on core-8-6-branch, in main, 9.0, 8.6: Fix bb5b343fe1: font tests failed on systems without the traditional X fonts (e.g. FreeBSD). Backport the font test constraints from branch less_tests_constra
        • 💬 2026-09-27 serhiy.storchaka: Still reproducible with core-8-6-branch on FreeBSD 15.1 with only the DejaVu fonts installed: font.test and unixFont.test expected fonts such as "times", and were only constrained with failsOnUbuntu. In 9.0 and main they were fixed by the new font constraints of branch less_tests_constraints (761407ca83), which were not merged into 8.6. Backported to 5af8f676bc. The font tests now pass on FreeBSD and Linux.
        Full verdict

        CANNOT-TEST (FreeBSD). NOT-REPRODUCED on Linux.

        • Platform generic (FreeBSD/XFCE)
        • Tested main (wt build), Xvfb, no WM, on a system without Times New Roman/Tahoma. font, unixFont, canvText, textIndex, text, event, config and unixMenu tests all pass (0 failures).
        • Cause font-dependent tests (the details are only in the attachment). Main's tests seem to have become font-agnostic since 2020.
        • Prototype none
        • Suggested ticket comment The font-sensitive test files pass on main under Linux/Xvfb without the MS web fonts installed. Can this be re-checked on FreeBSD with current 8.6/9.0? Otherwise close as out of date.
        ⬜3eb2a8df642020Custom treeview header -background unused for aqua Ttk themeCANNOT-TEST (macOS)

        Open ticket 3eb2a8df64 · 88. Themed Tk · filed 2020-07-21

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested code reading only. In main, macosx/ttkMacOSXTheme.c DrawListHeader() (now shared by light and dark mode) still fills with the hardcoded listheaderActiveBG/listheaderInactiveBG in light mode and fills nothing in dark mode. TreeHeaderElementDraw does not look at any -background option.
        • Cause see above. This is by design: it follows the native NSTableView look (Marc Culler's comment).
        • Prototype none (the decision is whether the aqua theme should honour Heading -background at all)
        • Suggested ticket comment Still the case in main: DrawListHeader() ignores Heading -background. The maintainers have to decide whether aqua should honour it, or whether this is NOT-A-BUG because it follows the native look.
        ⏳685ac307272020Aqua: XCopyArea() broken for non-pixmap source (as used by Tktable)CANNOT-TEST (macOS). Probably FIXED in 9.0+ by the CGBitmapContext backing store (merge 1dfe4fe9, 2024-07-08, "cgimage_with_crossing", implementi…

        Open ticket 685ac30727 · 82. X11 Emulation · filed 2020-04-08

        ⏳ Since triage (updated 2026-11-11)

        Status now: Pending (Out of Date)

        • 💬 2026-10-08 serhiy.storchaka: Since 1dfe4fe90a (Tk 9.0), Tk on macOS draws each window into its own bitmap, and XGetImage()/XCopyArea() with a window source copy from that bitmap, as proposed in 3e9e82bcfe. So this looks fixed in 9.0 and later. 8.6 uses [view.layer renderInContext:] instead. Marc, Christopher, could you confirm with the TkImg capture script above?
        • 💬 2026-10-08 chrstphrchvz: This ticket/discussion is rather outdated and probably should have been closed already, but no one (myself included) got around to it. I am not caught up on changes from the past several months, but I think any current issues with XGetImage()/XCopyArea() in 9.0+ are for reasons other than ones discussed in this ticket (see e.g. f215a9fdaa). I will have to try the TkImg example again later.
        Full verdict

        CANNOT-TEST (macOS). Probably FIXED in 9.0+ by the CGBitmapContext backing store (merge 1dfe4fe9, 2024-07-08, "cgimage_with_crossing", implementing the approach of the closed 3e9e82bcfe). CreateCGImageFromDrawableRect() now reads a window's contents from TKContentView.tkLayerBitmapContext. 8.6 still uses [view.layer renderInContext:].

        • Platform Aqua
        • Tested source reading only (main, 9.0, 8.6 macosx/tkMacOSXImage.c)
        • Cause before 9.0, Aqua could not read back the contents of a window, so XCopyArea() from a window source did not work.
        • Prototype none (macOS)
        • Suggested ticket comment Since 1dfe4fe9 (9.0) Aqua windows draw into a CGBitmapContext, and XCopyArea() copies from it, so XCopyArea() with a window source should work now. Can someone with a Mac check Tktable overflow drawing with 9.0 or later?
        ⬜84053449682020Behavior of wm geometry for fullscreen toplevels on Aqua (and possibly other platforms)CANNOT-TEST (macOS). The dead code is still there on main and 8.6: WM_FULLSCREEN is only defined (tkMacOSXWm.h) and tested in UpdateGeometryInfo()/…

        Open ticket 8405344968 · 66. Aqua Window Operations · filed 2020-03-15

        Full verdict

        CANNOT-TEST (macOS). The dead code is still there on main and 8.6: WM_FULLSCREEN is only defined (tkMacOSXWm.h) and tested in UpdateGeometryInfo()/ParseGeometry() (tkMacOSXWm.c:4905, 5246) and is never set.

        • Platform Aqua
        • Tested source reading only
        • Cause macosx/tkMacOSXWm.c checks wmPtr->flags & WM_FULLSCREEN, but -fullscreen is tracked through [macWindow styleMask] & NSFullScreenWindowMask.
        • Prototype none (the reporter asks what the intended behaviour is, e.g. for Split View, so this is a design question and needs macOS testing)
        • Suggested ticket comment WM_FULLSCREEN is still never set on Aqua, so these two checks are dead code. Replacing them with the NSWindowStyleMaskFullScreen test is simple, but whether wm geometry should be ignored in Split View needs someone with a Mac to decide.
        ⬜f1534f9d532020update does not work properly in some casesCANNOT-TEST (macOS; no small repro. bll could not reproduce it with the reporter's application or tclkit)

        Open ticket f1534f9d53 · 66. Aqua Window Operations · filed 2020-02-20

        Full verdict

        CANNOT-TEST (macOS; no small repro. bll could not reproduce it with the reporter's application or tclkit)

        • Platform Aqua
        • Tested none
        • Cause unknown. It was probably a window being closed during update in the application's own code.
        • Prototype none
        • Suggested ticket comment This could not be reproduced (also by bll with the provided application), and there is no minimal script. Closing as works-for-me; please reopen with a small script if it still happens with 8.6.17 or 9.0.
        ⬜c98281bed32019after idle behavior changeCANNOT-TEST (macOS). NOT-REPRODUCED on X11: after idle scheduled from a timer without any GUI events fires normally (main, 8.6). The Aqua notifie…

        Open ticket c98281bed3 · 69. Events · filed 2019-11-18

        Full verdict

        CANNOT-TEST (macOS). NOT-REPRODUCED on X11: after idle scheduled from a timer without any GUI events fires normally (main, 8.6). The Aqua notifier in main still has the Heartbeat timer (macosx/tkMacOSXNotify.c) that Marc suspected.

        • Platform Aqua
        • Tested main, 8.6 on Xvfb, no WM (after idle rescheduled every 50 ms: fired 40 times in 2 s)
        • Cause probably the Aqua notifier setupProc/heartbeat in 8.6.10-rc (macOS only)
        • Prototype none
        • Suggested ticket comment This does not happen on X11. The reporter did not answer Marc's question and there is no small script. Can someone check on macOS with 8.6.17 or 9.0, or should we close it?
        ⬜17f427c7282019X11 collision name header if build with aqua for MacOSXCANNOT-TEST (macOS). Still present in main from the source: unix/Makefile.in install-headers copies $(AQUA_XLIB_HDRS) (xlib/X11/*.h) into `$(INCL…

        Open ticket 17f427c728 · 99. Other · filed 2019-07-01

        Full verdict

        CANNOT-TEST (macOS). Still present in main from the source: unix/Makefile.in install-headers copies $(AQUA_XLIB_HDRS) (xlib/X11/*.h) into $(INCLUDE_INSTALL_DIR)/X11.

        • Platform build (Aqua)
        • Tested source inspection only
        • Cause unix/Makefile.in install-headers target; tk.h and tkIntXlibDecls.h include <X11/...>.
        • Prototype none. Moving the headers (e.g. to include/tk/X11 or X11_tk as MacPorts does) changes the public include layout for extensions. That is a packaging/TIP-level decision.
        • Suggested ticket comment Still the case on main: Aqua builds install Tk's stub X11 headers into $prefix/include/X11. Relocating them changes how extensions include tk.h, so this needs a decision (TIP?) rather than a simple fix.
        ⬜1bf48767292019path items should accept empty value for option -stateCANNOT-TEST (tkpath is not part of Tk; TIP 510 was rejected/Deferred, and the tip-510 branch was last touched 2019-05-08)

        Open ticket 1bf4876729 · -- New Widgets · filed 2019-04-13

        Full verdict

        CANNOT-TEST (tkpath is not part of Tk; TIP 510 was rejected/Deferred, and the tip-510 branch was last touched 2019-05-08)

        • Platform generic
        • Tested none. No path/tkpath sources in main, 8.7, 9.0 or 9.1.
        • Cause n/a. This belongs to the external tkpath package.
        • Prototype none (the code is not in Tk)
        • Suggested ticket comment The path items (TIP 510) were never merged into Tk; the TIP is Deferred. This should go to the tkpath project. Closing here.
        ⬜2f0e8922fc2018Strange artifactsCANNOT-TEST (OpenBSD only, rare, under low memory; there is no reproducer and the reporter does not blame Tk)

        Open ticket 2f0e8922fc · 02. Appearance · filed 2018-02-24

        Full verdict

        CANNOT-TEST (OpenBSD only, rare, under low memory; there is no reproducer and the reporter does not blame Tk)

        • Platform X11 (OpenBSD)
        • Tested none
        • Cause unknown. Underlines drawn as blocks and a dot for the cursor suggest a corrupted GC (line width/dash) or a server-side problem under memory pressure.
        • Prototype none
        • Suggested ticket comment Without a reproducer we cannot act on this. The reporter only wanted it recorded. Closing as Works For Me; please reopen with a recipe if it shows up with a current Tk.
        ⬜25cb062bda2017Allow Tk to be run in a child thread on MacOSCANNOT-TEST (macOS)

        Open ticket 25cb062bda · 83. Mac OS X Build · filed 2017-12-01

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none
        • Cause AppKit may only be used from the main thread, so Tk on Aqua cannot run in a secondary thread. The later segfault in Tk_DestroyWindow comes from AppKit calling into the child thread's Tk from the main thread.
        • Prototype none (would need an architecture redesign that forwards all AppKit calls to the main thread).
        • Suggested ticket comment AppKit requires the main thread, so loading Tk in a secondary thread is unsupported on macOS. Suggest closing as Won't Fix, or turning it into an RFE, and documenting the restriction.
        ⬜4edd8de6322017tk_getSaveFile returns initial file as part of directoryCANNOT-TEST (macOS)

        Open ticket 4edd8de632 · 37. [tk_get*File] · filed 2017-11-24

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none. Kevin Walzer could not reproduce it on 8.6.7, and the Aqua file dialogs (macosx/tkMacOSXDialog.c) have been reworked several times since.
        • Cause unknown (probably NSSavePanel state reused across calls with -initialfile).
        • Prototype none
        • Suggested ticket comment This could not be reproduced on 8.6.7, and the macOS file dialog code has changed a lot since. Please retest with 8.6.16 or 9.0; otherwise we will close it as Works For Me.
        ⬜87ff0c234c2017titlebar is wrong color on Mac OS XCANNOT-TEST (macOS; no repro script, the reporter would not share code; macOS 10.12 / Tk 8.6.7)

        Open ticket 87ff0c234c · 08. [(label)frame] and [toplevel] · filed 2017-10-12

        Full verdict

        CANNOT-TEST (macOS; no repro script, the reporter would not share code; macOS 10.12 / Tk 8.6.7)

        • Platform Aqua
        • Tested none
        • Cause unknown
        • Prototype none
        • Suggested ticket comment No reproducer was provided and the report is for macOS 10.12 with Tk 8.6.7. Can you still reproduce it with a current 8.6 or 9.0 and a minimal script? Otherwise we will close this ticket.
        ⬜cd437608562016OSX menus don't work in tkconCANNOT-TEST (macOS)

        Open ticket cd43760856 · 11. Aqua Menus · filed 2016-04-11

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none
        • Cause unknown. The reporter saw an initial grab on 8.6.5. kevin_walzer could not reproduce it twice. The Aqua menu code has been rewritten a lot since 8.6.5.
        • Prototype none (macOS only)
        • Suggested ticket comment Cannot reproduce by the maintainer on 8.6.5, and the macOS menu code has changed a lot since then. Please retest with 8.6.17 or 9.0; otherwise we will close this.
        ⬜218561bf8d2016Tk8.6.5 - Starkit binary under OSX 10.9.5 / OSX 10.11.3 crashes when resizing main window while dragging with the mouseCANNOT-TEST (macOS)

        Open ticket 218561bf8d · 66. Aqua Window Operations · filed 2016-03-17

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none
        • Cause unknown (CGImageCreate invalid bits/pixel during resize on 8.6.5). The Aqua drawing code was rewritten after 8.6.5 (layer-based drawing in 8.6.10+).
        • Prototype none (macOS only)
        • Suggested ticket comment The macOS drawing code has been rewritten since 8.6.5. Please retest with a current 8.6/9.0 wish (not a starkit) and provide a plain script if it still crashes; otherwise we will close this.
        ⬜01ef6cad3320161000x slowdown in tk X11 widget drawing speed Mac OS X 10.10+CANNOT-TEST (XQuartz/macOS)

        Open ticket 01ef6cad33 · 67. Unix Window Operations · filed 2016-02-18

        Full verdict

        CANNOT-TEST (XQuartz/macOS)

        • Platform X11 (XQuartz on macOS)
        • Tested sanity check on Linux Xvfb: the 400-button script takes 28 ms (main) and 19 ms (8.6), so there is no generic X11 slowdown.
        • Cause XQuartz/macOS interaction (App Nap/occlusion); the reporter also filed it with XQuartz and Apple. Not a Tk bug on real X servers.
        • Prototype none
        • Suggested ticket comment Drawing is fast on Linux X11 (about 20-30 ms for the script), so this is specific to XQuartz on macOS 10.10+. Proposing to close as a third-party issue unless it can still be reproduced with a current XQuartz.
        ⬜486a466e632015OS X + Xquartz = [font actual zapfdingbats] fatalCANNOT-TEST (macOS/XQuartz). Probably FIXED by a workaround in 9.0.1 and 8.6.16.

        Open ticket 486a466e63 · 46. Unix Fonts · filed 2015-02-10

        Full verdict

        CANNOT-TEST (macOS/XQuartz). Probably FIXED by a workaround in 9.0.1 and 8.6.16.

        • Platform X11 on XQuartz (macOS)
        • Tested font actual zapfdingbats on Linux/Xvfb works on main, 8.6 and 8.5 (Xft path).
        • Cause the XQuartz server returns BadValue for X_OpenFont (XQuartz issue #216). Since d2f53df7a8 (2024-09-22, main/9.0), with the backport to 8.6 in 5c47c3c280 (2024-11-13), unix/tkUnixFont.c wraps every XLoadQueryFont in XLoadQueryFontNoXError, which installs a Tk error handler for BadValue/X_OpenFont (ticket 36e379c01b). That is exactly this crash.
        • Prototype none (already done).
        • Suggested ticket comment Tk now ignores the BadValue error from X_OpenFont on XQuartz (workaround for 36e379c01b, check-ins d2f53df7a8 / 5c47c3c280, in 8.6.16 and 9.0.1). Please retest; this ticket is probably a duplicate and can be closed.
        ⬜326ed7bd6f2015[ttk:PanedWindow]CANNOT-TEST. This is a transient visual artifact while dragging, and I found no reliable way to observe it in Xvfb. It is also essentially by desig…

        Open ticket 326ed7bd6f · 88. Themed Tk · filed 2015-02-04

        Full verdict

        CANNOT-TEST. This is a transient visual artifact while dragging, and I found no reliable way to observe it in Xvfb. It is also essentially by design, as jenglish explains.

        • Platform generic (X11, Windows)
        • Tested not measured.
        • Cause when a child window moves or resizes, the server repaints (background or bit gravity) before Tk handles ConfigureNotify and redraws. On a non-compositing display this gives brief "stepped" edges. The same effect shows with plain place, so it isn't specific to panedwindow.
        • Prototype none (fixing it would mean redesigning the redraw, e.g. double buffering the whole toplevel).
        • Suggested ticket comment This is the inherent latency between the window system moving child windows and Tk redrawing them. It is not specific to ttk::panedwindow (plain ''place'' shows it too). Suggest closing as won't fix unless someone proposes a concrete approach.
        ✅4face822b02015wm attributes -fullscreen w/dual monitor opens to fullscreen in the wrong displaypatchCANNOT-TEST for two monitors: the VM has one monitor, and changing the VirtualBox monitor count needs the VM to be powered off. The prototype compi…

        Open ticket 4face822b0 · 68. Win Window Operations · filed 2015-01-10

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Fixed)

        • 🔧 2026-10-03 check-in 94e8295d08 on win-fullscreen-monitor, in main, 9.0: Fix 4face822b0: wm attributes -fullscreen w/dual monitor opens to fullscreen in the wrong display. UpdateWrapper always used the primary screen; use the monit
        • 🔧 2026-10-08 check-in 52a9c1193b on trunk, in main, 9.0, 8.6: Fix for 4face822b0: wm attributes -fullscreen w/dual monitor opens to fullscreen in the wrong display.
        • 🔧 2026-10-08 check-in c30bc468a9 on core-9-0-branch, in main, 9.0: Fix for 4face822b0: wm attributes -fullscreen w/dual monitor opens to fullscreen in the wrong display.
        • 🔧 2026-10-08 check-in 5a3ac1792d on core-8-6-branch, in main, 9.0, 8.6: Fix for 4face822b0: wm attributes -fullscreen w/dual monitor opens to fullscreen in the wrong display.
        • 💬 2026-10-03 serhiy.storchaka: UpdateWrapper (win/tkWinWm.c) always used the origin and size of the primary screen for full screen mode. Proposed fix in branch win-fullscreen-monitor: use the monitor nearest the window (MonitorFromRect, GetMonitorInfo). Tested on Windows 11 with two monitors: a normal, zoomed or not yet mapped window on the second monitor now goes full screen on the second monitor, and returns to its previous geometry or zoomed…
        • 💬 2026-10-08 nemethi: Many thanks, Serhiy! Merged into main, core-9-0-branch, and core-8-6-branch by commits 52a9c119, c30bc468, and 5a3ac179.
        Full verdict

        CANNOT-TEST for two monitors: the VM has one monitor, and changing the VirtualBox monitor count needs the VM to be powered off. The prototype compiles and does not change single-monitor behaviour.

        • Platform Windows
        • Tested code reading only. There is no Windows or MinGW here, so the prototype has not been compiled.
        • Cause win/tkWinWm.c:UpdateWrapper, the WM_FULLSCREEN branch of the geometry computation.
        • Prototype 4face822b0.patch (27 changed lines, uncompiled). It uses MonitorFromRect on the window's current rectangle (MONITOR_DEFAULTTONEAREST) plus GetMonitorInfoW rcMonitor, and falls back to the old behaviour. Repro script: win/4face822b0.tcl (needs two monitors; prints VERDICT).
        • Suggested ticket comment Still hardcoded in UpdateWrapper (fullscreen always at 0,0 with the primary screen size). A prototype that uses MonitorFromRect/GetMonitorInfo for the window's current monitor is prepared and needs testing on a multi-monitor Windows system.
        Windows VM, run 3
        • Platform Windows
        • Verdict CANNOT-TEST for two monitors: the VM has one monitor, and changing the VirtualBox monitor count needs the VM to be powered off. The prototype compiles and does not change single-monitor behaviour.
        • Tested 4face822b0.patch built in tk-wt91 (no warnings). work-win/fs.tcl on ref91 and on the prototype: normal→fullscreen→normal gives 1023x768+0+0 then back to 300x200+120+90; zoomed→fullscreen→normal gives 1023x768+0+0 then back to zoomed. Both builds give the same results; the only difference is the zoomed position, which comes from the 557441302b patch in the same build. wm.test, winWm.test, winDialog.test, menu.test and winMenu.test: 1096 tests, 0 failed.
        • Cause win/tkWinWm.c UpdateWrapper, WM_FULLSCREEN branch (x=y=0, WidthOfScreen/HeightOfScreen)
        • Prototype 4face822b0.patch (unchanged, 27 lines; compiles cleanly with MSVC)
        • Suggested ticket comment A fix using MonitorFromRect/GetMonitorInfo is ready. It is verified on a single monitor, where behaviour does not change, but still needs someone with two monitors to test it.
        ⬜ef83a5d7222014Mac OS X Tcl/Tk FilaDialog generates constraints issueCANNOT-TEST (macOS)

        Open ticket ef83a5d722 · 36. [tk_chooseDirectory] · filed 2014-11-26

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none (no macOS)
        • Cause The "Unable to simultaneously satisfy constraints" messages come from Apple's own open panel views (FI_TBrowserBackgroundView, FI_TListScrollView are Finder-internal classes), not from Tk-created views. This was a known Yosemite system log noise and is harmless.
        • Prototype none (not in Tk code)
        • Suggested ticket comment The conflicting constraints are between Apple-internal views of the NSOpenPanel, not Tk views, and the dialog still works. This was Yosemite-era system log noise. Closing unless someone can reproduce it on a current macOS.
        ⬜4f84af7b4c2014[tk fontchooser] does not work on Mac OS X if no initial font configuredCANNOT-TEST (macOS); code on main still looks affected

        Open ticket 4f84af7b4c · 33. Generic Dialog Support · filed 2014-11-12

        Full verdict

        CANNOT-TEST (macOS); code on main still looks affected

        • Platform Aqua
        • Tested code reading only (main macosx/tkMacOSXDialog.c)
        • Cause -[TKApplication changeFont:] does [fm convertFont:fontPanelFont], and fontPanelFont is a static that starts as nil. Nothing initialises it unless -font is configured with a non-empty value. convertFont: on nil returns nil, so no selection is recorded, FontchooserEvent gets a NULL font description, and neither -command nor <<TkFontchooserFontChanged>> fires.
        • Prototype none (macOS-only, untestable here). Probable fix: in changeFont:, fall back to [fm selectedFont] or [NSFont systemFontOfSize:0] when fontPanelFont is nil. Another option is to set a default fontPanelFont in FontchooserShowCmd.
        • Suggested ticket comment On main, changeFont: still converts a nil fontPanelFont when no -font was configured, so the callback is never fired. The fix is probably to fall back to [fm selectedFont] or the system font. This needs a macOS tester.
        ⬜c18c36f80c2014Cocoa Tk can segfault when embedded Tcl interpreters are created and destroyed in the same app bundleCANNOT-TEST (macOS); probably FIXED (Marc Culler's TkpInit rewrite d71d6703/ebd79f48, 2017-11, could not reproduce with python -j3/-j10)

        Open ticket c18c36f80c · 66. Aqua Window Operations · filed 2014-10-30

        Full verdict

        CANNOT-TEST (macOS); probably FIXED (Marc Culler's TkpInit rewrite d71d6703/ebd79f48, 2017-11, could not reproduce with python -j3/-j10)

        • Platform Aqua
        • Tested code reading only
        • Cause The Apple event handlers use NSApp._eventInterp, which is set once in -[TKApplication _setup:] (TkpInit runs its init only once). On main, nothing resets _eventInterp to NULL when that interp is deleted. A dangling pointer is therefore still possible in theory if the first interp is deleted and a later interp receives Apple events.
        • Prototype none (macOS)
        • Suggested ticket comment Marc could not reproduce this after the 2017 TkpInit rewrite. _eventInterp is still never cleared when its interp is deleted, though. Registering a Tcl_CallWhenDeleted handler that NULLs it (or re-points it to the next interp in TkpInit) would close the remaining hole.
        ⬜ec7fa6bb5c2013Column and Row resize cursors could use better cursorsCANNOT-TEST (RFE, Windows-only)

        Open ticket ec7fa6bb5c · 88. Themed Tk · filed 2013-08-20

        Full verdict

        CANNOT-TEST (RFE, Windows-only)

        • Platform Windows
        • Tested code reading only. ttk::treeview uses [ttk::cursor hresize] (library/ttk/cursors.tcl), which maps to size_we (IDC_SIZEWE) on Windows.
        • Cause Windows has no stock IDC_* column-resize ("split") cursor. The one Explorer uses lives in comctl32 resources. Supporting it would mean shipping a .cur resource in Tk and adding a new symbolic cursor.
        • Prototype none (RFE)
        • Suggested ticket comment Windows provides no stock IDC_ cursor for column resizing, so this needs a cursor resource bundled with Tk. Reclassifying as a feature request.
        ⬜3612814fff2013font configure causes crashCANNOT-TEST (macOS)

        Open ticket 3612814fff · 11. Aqua Menus · filed 2013-05-07

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested the ticket script runs without problems on X11 with main and 8.6.18 (the menu code is platform-specific, so this says nothing about Aqua). The Aqua menu code has been rewritten several times since 8.5.13.
        • Cause unknown (crash in Aqua TkpConfigureMenuEntry, called from MenuWorldChanged)
        • Prototype none
        • Suggested ticket comment Can someone with macOS check whether this still crashes with 8.6.17+ or 9.0? The Aqua menu code has changed a lot since 8.5.13.
        ⬜3610493fff2013Keyboard is ignored with SCIMCANNOT-TEST (SCIM is unmaintained and not installed; serhiy did not reproduce it with ibus on 2026-09-19)

        Open ticket 3610493fff · 69. Events · filed 2013-04-10

        Full verdict

        CANNOT-TEST (SCIM is unmaintained and not installed; serhiy did not reproduce it with ibus on 2026-09-19)

        • Platform X11
        • Tested none new (related ticket a7729b3d58 re-checked today with the built-in Xlib IM, see below)
        • Cause probably SCIM's XIM server leaving the IC in a filtering state (XFilterEvent always True), as in a7729b3d58.
        • Prototype none
        • Suggested ticket comment Doesn't reproduce with ibus. SCIM is no longer maintained. Close as a SCIM issue (same as a7729b3d58).
        ⬜3603738fff2013menus can't post if application is not in foregroundCANNOT-TEST (macOS)

        Open ticket 3603738fff · 11. Aqua Menus · filed 2013-02-07

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested code reading only
        • Cause The 2013 code needed [NSApp keyWindow]. TkpPostMenu (macosx/tkMacOSXMenu.c) no longer does: it uses popUpMenuPositioningItem:atLocation:inView:nil (TIP 533, 4b28f74572, 2019), with the appearance taken from the parent toplevel's view. The failure path is probably gone.
        • Prototype none
        • Suggested ticket comment The macOS menu posting code was rewritten (TIP 533, 4b28f74572) and no longer depends on the key window. Can someone with macOS check whether tk_popup on a background Wish still raises an error?
        ⏳3593334fff2012checkbutton bitmap via config not highlighted on AquaCANNOT-TEST (macOS). Not generic: on X11, a checkbutton with -bitmap set at creation and one set later by configure get the same size (48x31) on ma…

        Open ticket 3593334fff · 03. [*button] and [label] · filed 2012-12-06

        ⏳ Since triage (updated 2026-11-11)

        Status now: Pending (Out of Date)

        • 💬 2026-10-08 serhiy.storchaka: Since 6d9ba34985 (Tk 8.6.4), buttons on macOS are no longer native NSButton controls whose type and flags were changed on configure. They are drawn with HITheme, and the button kind is recomputed on every configure, so a checkbutton or radiobutton configured with -bitmap later should look the same as one created with it. So this looks fixed. Kevin, could you confirm on macOS and close it as Out of Date?
        Full verdict

        CANNOT-TEST (macOS). Not generic: on X11, a checkbutton with -bitmap set at creation and one set later by configure get the same size (48x31) on main and 8.6.

        • Platform Aqua
        • Tested main, 8.6 on X11 (no difference)
        • Cause unknown. The 8.5.13-era button code (macosx/tkMacOSXButton.c) has been rewritten since. TkpComputeButtonGeometry recomputes btnkind on every configure, so the create/configure paths look symmetric now.
        • Prototype none
        • Suggested ticket comment Reported against 8.5.13; the Aqua button code has been rewritten since. Can someone with macOS check whether this still happens with 8.6.17 or 9.0?
        ⏳3587262fff2012Width of menubutton too short on aqua with textvariableCANNOT-TEST (macOS)

        Open ticket 3587262fff · 03. [*button] and [label] · filed 2012-11-14

        ⏳ Since triage (updated 2026-11-11)

        Status now: Pending (Out of Date)

        • 💬 2026-10-08 serhiy.storchaka: Since 6d9ba34985 (Tk 8.6.4) and the menubutton geometry rewrite for 70e531918e (Tk 8.6.10), the size of a menubutton on macOS is computed from the text layout in the current font whenever the text variable changes. So this looks fixed. Could someone with a Mac confirm with the attached script?
        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested X11 main, 8.6, 8.5: menubutton -font/-textvariable -indicatoron 1 gets the right reqwidth, and it grows correctly when the variable changes. So this is not generic.
        • Cause not known. The Aqua menubutton geometry code (macosx/tkMacOSXMenubutton.c TkpComputeMenuButtonGeometry) has been rewritten since 2012 (4b28f74572, 2019-02-02, TIP 533 / 70e531918e), so it may already be fixed.
        • Prototype none (macOS only)
        • Suggested ticket comment Cannot reproduce on X11. The Aqua menubutton geometry code was rewritten in 2019 (TIP 533, 70e531918e). Can someone on macOS check whether this still happens with 8.6.17+/9.0?
        ⬜3580194fff2012Width too narrow on MacOS XCANNOT-TEST (macOS)

        Open ticket 3580194fff · 03. [*button] and [label] · filed 2012-10-25

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none (Aqua only). Read the code only.
        • Cause in main, macosx/tkMacOSXMenubutton.c computes text-only width as width*avgWidth + 2*padX, then adds LEFT_INSET + RIGHT_INSET + 2*inset (from 4b28f74572, 2019-02-02). In 2012 the code used just width*avgWidth. The insets that were missing then are probably what the reporter saw, so this may be fixed.
        • Prototype none (macOS only)
        • Suggested ticket comment Since 2019 (TIP 533 / 70e531918e) the Aqua menubutton adds the native insets to the -width in characters. Please retest on macOS with a current 8.6/9.0.
        ⏳3580159fff2012indicatoron ignored on MacOS XCANNOT-TEST (macOS). The code suggests it is fixed.

        Open ticket 3580159fff · 03. [*button] and [label] · filed 2012-10-25

        ⏳ Since triage (updated 2026-11-11)

        Status now: Pending (Out of Date)

        • 💬 2026-10-08 serhiy.storchaka: Since 6d9ba34985 (Tk 8.6.4), a checkbutton or radiobutton with -indicatoron 0 and no image is drawn by Tk, not as a native check box, and is raised or sunken depending on its state. With an image it is drawn as a bevel button, also without the indicator. So this looks fixed. Kevin, could you confirm on macOS and close it as Out of Date?
        Full verdict

        CANNOT-TEST (macOS). The code suggests it is fixed.

        • Platform Aqua
        • Tested none (Aqua only). Read the code only.
        • Cause in main and 8.6, macosx/tkMacOSXButton.c ButtonBackgroundDrawCB draws a checkbutton/radiobutton natively only when indicatorOn || hasImageOrBitmap. Otherwise it draws it from scratch as a raised/sunken button, so -indicatoron 0 should now be honoured. This came from culler's geometry cleanup 6cfd57f032 (2019-01-18); the current form of the line is 991b8f0eda (2019-04-16).
        • Prototype none
        • Suggested ticket comment Since the 2019 Aqua button rework (6cfd57f032), a check/radiobutton with -indicatoron 0 and no image is drawn non-natively, without the checkbox. Probably fixed; needs confirmation on macOS.
        ⬜3573447fff2012[tk busy] causes crash on OS X/CocoaCANNOT-TEST (macOS)

        Open ticket 3573447fff · 66. Aqua Window Operations · filed 2012-10-01

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested X11 main and 8.6: the ticket script, looped 5x (button invoke under tk busy hold/forget, then post/unpost/select in the combobox). No hang, no crash.
        • Cause not known. It is specific to Aqua and is probably an interaction between the nested update event loop and the combobox popdown.
        • Prototype none
        • Suggested ticket comment Not reproducible on X11. Needs retesting on macOS with current 9.0/8.6.
        ✅3348994fff2011Freeze for Dialogs Created from ShortcutCANNOT-TEST (macOS; the attached repro script was deleted in 2012, and wordtech could not reproduce with a reduced example)

        Open ticket 3348994fff · 11. Aqua Menus · filed 2011-07-01

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Out of Date)

        • 💬 2026-10-08 serhiy.storchaka: The example was deleted, and the Cocoa event loop integration was fixed in [3028676] and reworked in 8.6.10. Also, since 33b3af1cea, menu actions, including those invoked by Command-key shortcuts, are run as idle tasks, so a dialog opened from them no longer runs inside AppKit's menu handling. Please open a new ticket if a current Tk still hangs.
        Full verdict

        CANNOT-TEST (macOS; the attached repro script was deleted in 2012, and wordtech could not reproduce with a reduced example)

        • Platform Aqua
        • Tested none
        • Cause unknown (the ticket suggests it is related to [3028676], Cocoa event loop during menu-shortcut handling)
        • Prototype none
        • Suggested ticket comment No reproducer is left (the attachment was deleted) and the Cocoa port has been largely rewritten since 8.6b1. Can someone still reproduce this with 8.6.17 or 9.0? Otherwise we will close it.
        ⬜3170714fff2011Menubutton and -direction flush does not work properly withCANNOT-TEST (macOS; the report is specific to the --enable-aqua=carbon build, which was removed in 8.6. Probably obsolete)

        Open ticket 3170714fff · 11. Aqua Menus · filed 2011-02-02

        Full verdict

        CANNOT-TEST (macOS; the report is specific to the --enable-aqua=carbon build, which was removed in 8.6. Probably obsolete)

        • Platform Aqua
        • Tested none (X11 menubutton code path is unrelated)
        • Cause Carbon event handling in the old Carbon menu code (reporter says it broke in 8.5.7)
        • Prototype none (Carbon backend no longer exists)
        • Suggested ticket comment The Carbon backend was removed in Tk 8.6. Can someone with macOS check whether this still happens with the Cocoa build? Otherwise it should be closed as out of date.
        ⬜3033786fff2010Cannot embed Tk in a Cocoa ApplicationCANNOT-TEST (macOS; this is really an RFE. Maintainers (das, kevin_walzer) said it is a non-goal)

        Open ticket 3033786fff · 83. Mac OS X Build · filed 2010-07-23

        Full verdict

        CANNOT-TEST (macOS; this is really an RFE. Maintainers (das, kevin_walzer) said it is a non-goal)

        • Platform Aqua
        • Tested none
        • Cause Tk-Cocoa subclasses NSApplication (TKApplication), which conflicts with a host application's NSApplication
        • Prototype none (a redesign)
        • Suggested ticket comment Embedding Tk into a foreign Cocoa application is not supported. Suggest closing as Won't Fix, perhaps with a note in the frame/toplevel -use/-container docs that embedding is not available on Aqua.
        ⬜3028365fff2010Default LIBDIR not on Tk's auto_pathCANNOT-TEST (macOS framework/Wish.app layout)

        Open ticket 3028365fff · 83. Mac OS X Build · filed 2010-07-12

        Full verdict

        CANNOT-TEST (macOS framework/Wish.app layout)

        • Platform Aqua (framework build)
        • Tested none
        • Cause Wish.app's info nameofexecutable is inside the framework, so Tcl's <exe>/../lib auto_path entry is never /usr/local/lib (see macosx/tkMacOSXInit.c auto_path setup)
        • Prototype none
        • Suggested ticket comment Needs checking on macOS with a current framework build. If it still happens, macosx/tkMacOSXInit.c could add the configured libdir (or the path of the wish symlink's ../lib) to auto_path.
        ⬜3002484fff2010Hard crash in wish on OSX when no windowserverCANNOT-TEST (macOS)

        Open ticket 3002484fff · 83. Mac OS X Build · filed 2010-05-16

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none. Code check: macosx/ still has no window-server/session check (no CGSessionCopyCurrentDictionary or similar) before NSApp/window creation.
        • Cause TkpInit (macosx/tkMacOSXInit.c) starts AppKit unconditionally. Without a window server session, creating the first NSWindow raises an uncaught NSException.
        • Prototype none (macOS only). Possible fix: in TkpInit, return TCL_ERROR ("no window server available") when CGSessionCopyCurrentDictionary() returns NULL. CPython's tkinter tests use the same check.
        • Suggested ticket comment Still no check for window server availability in the Aqua init code. A check with CGSessionCopyCurrentDictionary() in TkpInit that returns a Tcl error would avoid the abort.
        ⬜2976019fff2010Deployment executables go bus errorCANNOT-TEST (PPC/OS X 10.5 Deployment build of 8.6b1; platform long unsupported)

        Open ticket 2976019fff · 83. Mac OS X Build · filed 2010-03-24

        Full verdict

        CANNOT-TEST (PPC/OS X 10.5 Deployment build of 8.6b1; platform long unsupported)

        • Platform build (macOS PPC, OS X 10.5)
        • Tested none (no macOS/PPC)
        • Cause unknown; crash in TkFindStateNumObj -> Tcl_NewStringObj(strlen of bad pointer) at startup, only in optimized (Deployment) builds on PPC — likely compiler/ABI issue of that era.
        • Prototype none (no platform)
        • Suggested ticket comment PPC and OS X 10.5 are no longer supported by Tk 8.6/9.x, and no similar crash has been reported on current macOS. Closing as out of date; please reopen with a current build if it recurs.
        ⬜2974003fff2010winfo rootx/rooty broken on AquaCANNOT-TEST (macOS; Carbon Tk 8.4 is obsolete; already not reproducible by Kevin Walzer in 2011 on 8.4.19 and Cocoa 8.6)

        Open ticket 2974003fff · 66. Aqua Window Operations · filed 2010-03-21

        Full verdict

        CANNOT-TEST (macOS; Carbon Tk 8.4 is obsolete; already not reproducible by Kevin Walzer in 2011 on 8.4.19 and Cocoa 8.6)

        • Platform Aqua (Carbon 8.4.19)
        • Tested none (no macOS)
        • Cause n/a
        • Prototype none
        • Suggested ticket comment Carbon-based Tk 8.4 is obsolete; the maintainer could not reproduce it in 2011 with 8.4.19 or Cocoa 8.6. Closing as out of date.
        ⬜2969401fff2010Coordinates reported wrongly on ppcCANNOT-TEST (macOS 10.3/PPC, Carbon Tk; platform unsupported)

        Open ticket 2969401fff · 62. Geometry Functions · filed 2010-03-12

        Full verdict

        CANNOT-TEST (macOS 10.3/PPC, Carbon Tk; platform unsupported)

        • Platform Aqua (PPC, OS X 10.3)
        • Tested none (no macOS)
        • Cause unknown (Carbon-era window-move tracking)
        • Prototype none
        • Suggested ticket comment Reported for Carbon Tk on OS X 10.3/PPC, which is no longer supported. Closing as out of date; please reopen if it still happens with Tk 8.6/9.0 on current macOS.
        ✅2933099fff2010Cocoa Tk wraps text when no wrapping specifiedCANNOT-TEST (macOS; repro needs Tix on Cocoa Tk 8.5.8/8.6b1; macOS text drawing was rewritten since — CoreText in 8.6)

        Open ticket 2933099fff · 02. Appearance · filed 2010-01-15

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Duplicate)

        • 💬 2026-10-08 serhiy.storchaka: The wrapping was done by Tk, not by Cocoa. Tk_MeasureChars on macOS rounded the text width to the nearest pixel (77 for 77.4), so when the text was later laid out with that width as the limit, the last character did not fit and was wrapped. This was fixed in d6cbedac6d for [2857300] (Tk 8.6b2): the width is now rounded up.
        Full verdict

        CANNOT-TEST (macOS; repro needs Tix on Cocoa Tk 8.5.8/8.6b1; macOS text drawing was rewritten since — CoreText in 8.6)

        • Platform Aqua
        • Tested none (no macOS)
        • Cause likely old ATSUI/NSString drawing in the early Cocoa port wrapping when the given width was 1px short; code replaced by CoreText-based tkMacOSXFont.c.
        • Prototype none
        • Suggested ticket comment The macOS text drawing code has been rewritten (CoreText) since 8.6b1. Can anyone reproduce this with Tk 8.6.1x/9.0 (with Tix or a pure Tk example)? Otherwise we will close it as out of date.
        ⬜2908379fff2009ttk:* widgets - selection text illegible on Mac OS XCANNOT-TEST (macOS), probably FIXED

        Open ticket 2908379fff · 88. Themed Tk · filed 2009-12-03

        Full verdict

        CANNOT-TEST (macOS), probably FIXED

        • Platform Aqua
        • Tested none (no macOS). Code only: library/ttk/aquaTheme.tcl now sets -selectforeground systemSelectedTextColor / -selectbackground systemSelectedTextBackgroundColor for all ttk widgets (a8a59c2819, 2019-10-25), and Treeview selected foreground is systemAlternateSelectedControlTextColor (setTreeviewAndListboxSelectColors). das's 2009 suggestion is what is implemented now.
        • Cause 2009 aqua theme had no selected-text foreground color.
        • Prototype none (macOS-only, apparently already done)
        • Suggested ticket comment The aqua theme now uses systemSelectedTextColor/systemSelectedTextBackgroundColor for ttk selections (and systemAlternateSelectedControlTextColor for Treeview). Can someone with macOS confirm and close?
        ⬜2824538fff2009Need to track opaque content in textured windowsCANNOT-TEST (macOS)

        Open ticket 2824538fff · 66. Aqua Window Operations · filed 2009-07-20

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none. Code: macosx/tkMacOSXWm.c still has the 2009 workaround [window setMovableByWindowBackground:NO] for textured/HUD windows ("Workaround for [Bug 2824538]"). Ticket was left open by das only as a reminder for the long-term opaque-region tracking, which is an RFE-sized redesign.
        • Cause Tk does not expose an NSView hierarchy marking opaque subareas of a textured window.
        • Prototype none (macOS only, redesign)
        • Suggested ticket comment The workaround (textured windows not movable by background) is still in place; proper opaque-content tracking would be a redesign. Suggest closing, or converting to a Feature Request.
        ⬜2806562fff2009tk_chooseDirectory on Mac freezes when accessing Samba shareCANNOT-TEST (macOS; the native NSOpenPanel is also completely different from the 2009 Carbon code)

        Open ticket 2806562fff · 36. [tk_chooseDirectory] · filed 2009-06-15

        Full verdict

        CANNOT-TEST (macOS; the native NSOpenPanel is also completely different from the 2009 Carbon code)

        • Platform Aqua
        • Tested none
        • Cause unknown (Carbon-era Navigation Services dialog, since replaced by NSOpenPanel)
        • Prototype none
        • Suggested ticket comment This was reported against the Carbon dialog. Tk 8.6+/9 uses NSOpenPanel. Can someone on macOS check whether it still happens?
        ✅2534237fff2009Tk not linking correctly on netbsdCANNOT-TEST (NetBSD). Half of the patch is already applied; the other half is not.

        Open ticket 2534237fff · 84. Unix Build · filed 2009-01-25

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Out of Date)

        • 💬 2026-09-27 serhiy.storchaka: Out of date. The Makefile part is applied: X11_LIB_SWITCHES puts $(XFT_LIBS) before @XLIBSW@. And configure adds the X library directory to LIB_RUNTIME_DIR, so it is in the rpath. On NetBSD 10.1, libX11 is found at run time with Tk 8.6 and 9.1, with and without Xft.
        Full verdict

        CANNOT-TEST (NetBSD). Half of the patch is already applied; the other half is not.

        • Platform build
        • Tested source inspection only. The attachment rpath.diff (artifact 10bbae8e9a) was read from the repository.
        • Cause The Makefile.in part (X11_LIB_SWITCHES = $(XFT_LIBS) @XLIBSW@) was committed in 6d03bfe193 on 2009-11-19 ("Fix library order in X11_LIB_SWITCHES"). The tcl.m4 part (SC_PATH_X adding -Wl,-rpath,$x_libraries to XLIBSW on NetBSD) was never applied. Current NetBSD-* in tcl.m4 only puts an rpath for LIB_RUNTIME_DIR.
        • Prototype none. It would be about 7 lines in unix/tcl.m4 plus regenerating configure, but it cannot be verified without NetBSD. pkgsrc normally handles X11 rpaths with its wrappers.
        • Suggested ticket comment The library order in X11_LIB_SWITCHES was fixed in 6d03bfe193. The tcl.m4 part (rpath for $x_libraries on NetBSD) was never applied. Is it still needed with current NetBSD/pkgsrc? If nobody can confirm, we propose closing.
        ⬜2184632fff2008postcascade <index> does not workCANNOT-TEST (macOS)

        Open ticket 2184632fff · 11. Aqua Menus · filed 2008-10-21

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested X11 main, 8.6 as an analogy. .mbar postcascade 0 on the menubar master does nothing on X11 as well, because the visible menubar is the clone .#mbar. .#mbar postcascade 0 does post the File menu. On Aqua the menubar is a native NSMenu, and TkpPostMenu/postcascade cannot open it.
        • Cause on Aqua, postcascade on a native menubar entry is not implemented. On X11 the command acts on the unmapped master, not on the clone.
        • Prototype none (macOS; supporting it would mean forwarding postcascade from the master to its menubar clone or native menu, which is a design change)
        • Suggested ticket comment Still open on Aqua (not testable here). On X11, [.mbar postcascade] also acts on the unmapped master menu. Only the menubar clone (.#mbar) posts.
        ⬜2166563fff2008test case main-3.1 hangs displaying to Exceed x serverCANNOT-TEST (Exceed X server/Solaris not available)

        Open ticket 2166563fff · 86. Test Tools · filed 2008-10-14

        Full verdict

        CANNOT-TEST (Exceed X server/Solaris not available)

        • Platform X11 (Solaris + Hummingbird Exceed)
        • Tested main, 8.6 on Xvfb. wish -help exits right away even with stdin an open pipe. main.test passes, but main-3.1 is skipped on main, where it is now constrained !withXft: "Xft with bidi support ... can cause hangs" (058e738f9f, 2026-03-27). This build has --disable-bidi.
        • Cause unknown. The "% %" in the result suggests the child reached the interactive prompt instead of exiting after the -help error.
        • Prototype none
        • Suggested ticket comment Cannot reproduce on Xvfb with 8.6 or main. The 2008 Exceed setup is no longer testable. Note that main-3.1 is now skipped for Xft+bidi builds because of a similar hang, so that should be investigated separately.
        ⬜2100553fff2008Panic in ImgPhotoGetpatchCANNOT-TEST (the dangling pointer cannot be observed without a debug allocator). The prototype compiles and does not change behaviour.

        Open ticket 2100553fff · 82. X11 Emulation · filed 2008-09-08

        Full verdict

        CANNOT-TEST (the dangling pointer cannot be observed without a debug allocator). The prototype compiles and does not change behaviour.

        • Platform Windows
        • Tested code reading only. Repro script: win/2100553fff.tcl. It posts WM_DISPLAYCHANGE with a fake depth through PowerShell, then compares winfo visualid of the old and new windows and creates a photo in an old window. The verdict is written to 2100553fff.out.
        • Cause (suspected): win/tkWinX.c TkWinDisplayChanged frees and reallocates screen->root_visual on every depth change. Existing windows keep the old, dangling Visual*. XVisualIDFromVisual() on it later returns garbage, xlib/xutil.c XGetVisualInfo does not match, and TkImgPhotoGet panics. The default colormap is freed the same way while windows still reference it (not addressed).
        • Prototype 2100553fff.patch (28 lines, untested and not compiled). It updates root_visual in place and allocates it only once.
        • Suggested ticket comment TkWinDisplayChanged frees screen->root_visual while existing windows still point to it. The dangling visual is the likely cause of the panic after an RDP depth change. Prototype: update the visual in place (needs testing on Windows).
        Windows VM, run 4
        • Platform Windows
        • Verdict CANNOT-TEST (the dangling pointer cannot be observed without a debug allocator). The prototype compiles and does not change behaviour.
        • Tested win/2100553fff.tcl on ref91 and on the prototype build. Both print visualid 0x0 before and after 5 fake WM_DISPLAYCHANGE messages, with no panic. The freed Visual keeps its old contents under Tcl's allocator, so the repro cannot tell the builds apart. The color/photo tests above passed with the prototype.
        • Cause TkWinDisplayChanged frees and reallocates screen->root_visual (code reading, result-39)
        • Prototype 2100553fff.patch (28 lines, unchanged). Compiles cleanly with MSVC.
        • Suggested ticket comment Updating root_visual in place instead of freeing it avoids the dangling Visual* in existing windows. The prototype compiles and passes the tests, but the panic itself could not be reproduced (it needs a real depth change, e.g. RDP).
        Prototype verified 2026-10-24, Windows VM
        • Status CANNOT-TEST (the dangling Visual* cannot be observed without a real depth change or a debug allocator)
        • Applies to main clean
        • Repro not reproducible / not reproducible
        • Test none (needs a real display depth change, e.g. RDP)
        • Test file run B2: winWm 39/1 (winWm-13.1 is the test of the unapplied 1567354, expected), color 21/0, imgPhoto 248/0, winfo 76/0, frame 207/0. It compiles cleanly.
        • Patch changes none
        💬2031258fff2008problem with Aqua Tk_Text in Tk>8.4CANNOT-TEST (macOS). Probably obsolete: the ATSU code the report blames was removed by the Cocoa port (05bd0a2cb8, 2009-06-29, Tk 8.6b1/8.5.7+). Te…

        Open ticket 2031258fff · 18. [text] · filed 2008-07-29

        💬 Since triage (updated 2026-11-11)

        Status now: Open

        • 💬 2026-10-08 serhiy.storchaka: The ATSUI code is gone since the Cocoa port. But on macOS each display line of a long logical line is still measured with the rest of the segment as context (TK_LAYOUT_WITH_BASE_CHUNKS), and TkpMeasureCharsInContext builds a CoreText typesetter for that whole context. So laying out one long logical line is quadratic in its length. Could someone with a Mac time the script above with Tk 9.0.5 or 9.1.0, e.g. for 2000…
        Full verdict

        CANNOT-TEST (macOS). Probably obsolete: the ATSU code the report blames was removed by the Cocoa port (05bd0a2cb8, 2009-06-29, Tk 8.6b1/8.5.7+). Text measuring now uses CoreText (CTTypesetter).

        • Platform Aqua
        • Tested The script from the ticket on X11, main and 8.6: insert+update takes 4–16 ms, and scrolling or editing the long line takes about 20 ms. The generic text code is fine.
        • Cause The report points at ATSUBreakLine in the old Carbon TkAqua, which no longer exists.
        • Prototype none (macOS only)
        • Suggested ticket comment The ATSU-based code this refers to was replaced by the Cocoa/CoreText port in 2009. The script is fast on X11. Can someone on macOS confirm that it is fine with 8.6/9.x so the ticket can be closed?
        ⬜1977618fff2008tk_messageBox OS X no translationCANNOT-TEST (macOS). By code inspection it is still present on main.

        Open ticket 1977618fff · 38. [tk_messageBox] · filed 2008-05-29

        Full verdict

        CANNOT-TEST (macOS). By code inspection it is still present on main.

        • Platform Aqua
        • Tested none (source inspection only)
        • Cause macosx/tkMacOSXDialog.c alertButtonNames[] holds hard-coded English NSStrings (@"Yes", @"No", ...). They are passed to -[NSAlert addButtonWithTitle:] without NSLocalizedString or msgcat. The X11 msgbox.tcl uses [mc "&Yes"], and library/msgs/de.msg has these entries.
        • Prototype none (cannot build/test Aqua). Suggested fix: fetch the titles via ::msgcat::mc on "&Yes" etc. and strip the "&", as the Tk message box code does.
        • Suggested ticket comment Still present in the source: the Aqua tk_messageBox uses hard-coded English button titles (alertButtonNames in tkMacOSXDialog.c). They could be translated through the existing msgcat catalogs ("&Yes" -> "&Ja").
        ⬜1881923fff2008wm geometry with negative positions wrongCANNOT-TEST (macOS)

        Open ticket 1881923fff · 66. Aqua Window Operations · filed 2008-01-29

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none on macOS; X11 is not affected (the reporter says so too)
        • Cause reported against Carbon Tk 8.5.0. Current macosx/tkMacOSXWm.c (UpdateGeometryInfo/ParseGeometry) now computes WM_NEGATIVE_X/Y offsets from vRootWidth/vRootHeight, so it may already be fixed in Cocoa Tk. Needs checking on a Mac.
        • Prototype none (macOS only)
        • Suggested ticket comment The Carbon code this was reported against is gone. The Cocoa wm now handles WM_NEGATIVE_X/Y. Could someone with a Mac check whether wm geometry . -20-20 puts the window 20px from the bottom-right corner on 8.6/9.0?
        ⬜1881903fff2008tk_messageBox -icon on Mac wrongCANNOT-TEST (macOS)

        Open ticket 1881903fff · 38. [tk_messageBox] · filed 2008-01-29

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested code reading only
        • Cause the Carbon kAlert* mapping from the ticket no longer exists. The Cocoa macosx/tkMacOSXDialog.c already defaults to ICON_INFO (point 1 is fixed). It always calls setIcon: with the application icon and maps error/question to NSWarningAlertStyle and warning to NSCriticalAlertStyle. On modern macOS only the critical style shows a badge, so the icons still differ from other platforms by design.
        • Prototype none (macOS only, and the look is dictated by the OS)
        • Suggested ticket comment The Carbon implementation was replaced by NSAlert, and the default icon is now "info". The remaining differences come from NSAlert styles (only critical gets a badge). Suggest closing as Out of Date unless someone on current macOS disagrees.
        ⬜1847018fff2007X Error ...CANNOT-TEST (no reproducer; 8.5b3 / 8.4.12 on XQuartz Leopard and Debian Etch xorg 7.1)

        Open ticket 1847018fff · 67. Unix Window Operations · filed 2007-12-08

        Full verdict

        CANNOT-TEST (no reproducer; 8.5b3 / 8.4.12 on XQuartz Leopard and Debian Etch xorg 7.1)

        • Platform X11
        • Tested none. There is no script, and the only details are "Coccinella crashes at startup" with BadMatch on X_PolyFillRectangle.
        • Cause unknown. BadMatch on PolyFillRectangle usually means a GC/drawable depth mismatch (e.g. drawing on a pixmap or visual of a different depth), probably in the application or an extension (tktray/snack), or an X server bug of that era.
        • Prototype none
        • Suggested ticket comment No reproducer, and the report is against 8.4/8.5b3 on X servers from 2007. Closing as Out of Date. Please reopen with a script that triggers the BadMatch on a current Tk (run with -sync).
        ⬜1843539fff2007enter bindings after activation fire only after first moveCANNOT-TEST (macOS)

        Open ticket 1843539fff · 66. Aqua Window Operations · filed 2007-12-03

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none. The behaviour depends on Aqua (Carbon 8.4 at the time) mouse-tracking and activation events and is not testable on X11.
        • Cause the Carbon 8.4 event handling of that time. Cocoa Tk (tkMacOSXMouseEvent.c / tkMacOSXWindowEvent.c) was rewritten since.
        • Prototype none
        • Suggested ticket comment Reported against Carbon Tk 8.4. Could someone on macOS check whether <Enter> fires on click-activation without moving the mouse in 8.6/9.0? Otherwise close as Out of Date.
        🟢1751560fff2007make test failuresCANNOT-TEST (Solaris/CDE)

        Open ticket 1751560fff · 86. Test Tools · filed 2007-07-10

        🟢 Since triage (updated 2026-11-11)

        Status now: Open

        • 🔧 2026-09-30 check-in 7c7b6d7a2b on x11-wm-position-work-area, in main, 9.0: Fix the position reported by "wm geometry" on X11 when the window manager moves a toplevel from the requested position to the edge of the work area (_NET_WORKAR
        • 💬 2026-09-30 serhiy.storchaka: Tested with 9.1b1 on Linux/X11: * window.test does not hang with no window manager, fvwm, icewm, openbox, xfwm4, metacity and KWin. * font.test no longer has the message about expected failures. * unixWm-4.3 still fails with a panel at the top of the screen, e.g. under KWin and xfwm4, because wm geometry reports the requested position instead of the position the window manager moved the window to; see d1a711b500…
        Full verdict

        CANNOT-TEST (Solaris/CDE)

        • Platform build (Solaris 9, CDE/GNOME)
        • Tested main wt build on Xvfb without a WM. window.test 18/18 passed with no hang. font.test 292/0 failures. unixWm-4.x passed (other unixWm failures are the usual iconify tests with no WM). The unixWm-4.x tests were rewritten (+$X+$Y0 geometry). On X11, Y0 is still 0, so a WM panel at the top of the screen would still shift the result.
        • Cause environment-specific. The WM (a top panel or struts) moves the window, and the window.test hang was on CDE.
        • Prototype none
        • Suggested ticket comment The test suite has changed a lot since 8.4.15. window.test and font.test pass on current Linux/X11. unixWm tests depend on the WM (panels/struts) by nature. Closing as out of date. Please open a new ticket if you still see failures with a current Tk.
        ⏳1683795fff2007-padx obscures button text on AquaCANNOT-TEST (macOS)

        Open ticket 1683795fff · 03. [*button] and [label] · filed 2007-03-19

        ⏳ Since triage (updated 2026-11-11)

        Status now: Pending

        • 💬 2026-09-25 serhiy.storchaka: The Carbon code is gone. The current Cocoa button code adds room for the native push button bezel on top of the text and -padx, and places the text inside the padding, so button .b -text Foobazzulous -padx 5 should no longer clip the text. Can someone on macOS confirm with 8.6.18 or 9.0?
        • 💬 2026-09-25 kevin_walzer: Issue not observed in Tk 9.1 tip.
        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none. X11 is not affected, as the report says.
        • Cause unknown. The report is from the Carbon era. The current Cocoa button code (macosx/tkMacOSXButton.c) is completely different, so the old analysis probably no longer applies.
        • Prototype none (Aqua-only)
        • Suggested ticket comment The report is about Carbon Tk. Can someone with macOS check whether button .b -text Foobazzulous -padx 5 still clips the text with Cocoa Tk 8.6/9.0? If not, this can be closed.
        ⬜1622189fff2006XIM does not work with ATOK12 on Solaris8CANNOT-TEST (needs ATOK12 on Solaris 8/9, platforms unsupported since 2014; ibus and fcitx do not take the focus with their candidate window)

        Open ticket 1622189fff · 69. Events · filed 2006-12-26

        Full verdict

        CANNOT-TEST (needs ATOK12 on Solaris 8/9, platforms unsupported since 2014; ibus and fcitx do not take the focus with their candidate window)

        • Platform X11
        • Tested none (as serhiy's 2026-09-19 comment says, TkFocusKeyEvent still drops key events that arrive while there is no focus window)
        • Cause generic/tkFocus.c:TkFocusKeyEvent returns NULL for a KeyPress (keycode 0, the XIM commit) that arrives between FocusOut and FocusIn, so the committed text is lost.
        • Prototype none (no IM to test against. A possible fix: when focus is lost to an IM auxiliary window, send keycode-0 KeyPress events to the last focus window, the displayFocusPtr->focusWinPtr from before the FocusOut)
        • Suggested ticket comment Can't be tested without ATOK on Solaris. Closing as obsolete unless someone can reproduce it with a current input method whose candidate window takes the focus.
        ⬜1611821fff2006tk_getSaveFile -initialfile odd behaviorCANNOT-TEST (macOS). Code reading suggests it still exists.

        Open ticket 1611821fff · 37. [tk_get*File] · filed 2006-12-08

        Full verdict

        CANNOT-TEST (macOS). Code reading suggests it still exists.

        • Platform Aqua
        • Tested X11 (tkfbox), main, 8.6 and 8.4. A full path in -initialfile is shown as-is in the entry, the dialog stays in the current or -initialdir directory, and OK returns the full path. So X11 accepts it, but does not open the file's directory.
        • Cause macosx/tkMacOSXDialog.c (OPEN_INITFILE/SAVE_INITFILE) passes the whole -initialfile string to -[NSSavePanel setNameFieldStringValue:], which treats '/' as an illegal filename character.
        • Prototype none (Aqua-only and untestable). Suggested fix, following dkf: if -initialfile has a directory part, [file join $initialdir $initialfile], then use file dirname as the directory and file tail as the name. Do this in the Aqua code and, for consistency, in tkfbox.tcl/xmfbox.tcl. getOpenFile.n should also say what a path in -initialfile means.
        • Suggested ticket comment The Aqua code still passes -initialfile unchanged to setNameFieldStringValue:. We could split a directory part into the initial directory on all platforms, as dkf suggested.
        💬1591488fff2006test suite tests 10.2, 11.1 failedCANNOT-TEST (depends on the Exceed X server). NOT-REPRODUCED on Xvfb.

        Open ticket 1591488fff · 53. [selection] · filed 2006-11-06

        💬 Since triage (updated 2026-11-11)

        Status now: Open

        • 💬 2026-09-30 serhiy.storchaka: The same cause as [1719278]: Exceed requests the PRIMARY selection as well, and the extra calls of the handler show up in the results. Proposed fix in branch select-tests-private-selection (29a21fe034): select-10.2 and select-11.1 use a private selection; see [610188].
        Full verdict

        CANNOT-TEST (depends on the Exceed X server). NOT-REPRODUCED on Xvfb.

        • Platform X11 (Solaris 2.9 client, Hummingbird Exceed X server)
        • Tested select.test on main (wt build) and 8.6 tip, on Xvfb: 95 passed, 0 failed on both.
        • Cause X server behaviour. Under Exceed the handler was called twice for select-10.2 (a repeated conversion request), and for 11.1 the requestor got a "doesn't exist" error where "owner didn't respond" was expected. select-11.1 has since been marked failsOnCILinux (4a1e78ed7c, 2026-01-13), so it is timing-sensitive on other servers too.
        • Prototype none (no Tk bug identified)
        • Suggested ticket comment select-10.2 and select-11.1 pass on 8.6 and 9.x under Xvfb. The failures depended on the Exceed X server, and 11.1 is now marked failsOnCILinux. Closing as obsolete.
        ⬜1581490fff2006Tk button on Aqua differs between -text and -imageCANNOT-TEST (macOS)

        Open ticket 1581490fff · 02. Appearance · filed 2006-10-20

        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none
        • Cause Aqua button rendering picks a different native bezel style for image buttons than for text buttons (macosx/tkMacOSXButton.c). This is probably intentional.
        • Prototype none
        • Suggested ticket comment Needs checking on current macOS. The button code has been rewritten since 2006, and the choice of bezel style for image buttons is deliberate.
        ⬜1567292fff2006font size and tk scalingCANNOT-TEST (macOS). X11 behaves as the reporter expects from 8.4.

        Open ticket 1567292fff · 45. Aqua Fonts · filed 2006-09-28

        Full verdict

        CANNOT-TEST (macOS). X11 behaves as the reporter expects from 8.4.

        • Platform Aqua
        • Tested X11 main, 8.6 and 8.4 on Xvfb. With the ticket's canvas script, fonts follow the tk scaling in effect when they were created: widths 97/161/136 px for 14pt at 1.25, 12pt at 2.5 and 10pt at 2.5. A font that is still in use keeps its size, because of the cache. The 2011 comment's idiom font configure $f -size [font configure $f -size] does rescale named fonts on X11: 100 → 300 px going from scaling 1 to 3.
        • Cause on Aqua, point sizes map to pixels at 72 dpi regardless of tk scaling (macosx/tkMacOSXFont.c). That is by design since 8.5.
        • Prototype none
        • Suggested ticket comment On X11, fonts follow tk scaling and reconfiguring named fonts rescales them. On Aqua, font sizes deliberately ignore tk scaling. This needs a decision from an Aqua maintainer and is probably not a bug.
        ⏳1488195fff2006Solaris LP64 build incorrectly hardcodes /usr/openwin/libCANNOT-TEST (Solaris/SPARC 64-bit; no such platform). main's and 8.6's tcl.m4 still include /usr/openwin/lib, with no /64 variant, in the SC_PATH_X…

        Open ticket 1488195fff · 84. Unix Build · filed 2006-05-14

        ⏳ Since triage (updated 2026-11-11)

        Status now: Pending (Out of Date)

        • 💬 2026-09-25 serhiy.storchaka: Still true in current configure: the X library directory is added to the runtime path (-R on Solaris), so if /usr/openwin/lib is found, a 64-bit build gets the 32-bit directory. Workaround: --x-libraries=/usr/openwin/lib/64. OpenWindows is gone since Solaris 11 and this cannot be tested here, so this is likely out of date.
        Full verdict

        CANNOT-TEST (Solaris/SPARC 64-bit; no such platform). main's and 8.6's tcl.m4 still include /usr/openwin/lib, with no /64 variant, in the SC_PATH_X fallback search list, so the issue probably still applies there (only when X is found through that fallback).

        • Platform build (Solaris)
        • Tested source inspection (unix/tcl.m4 SC_PATH_X in main and 8.6)
        • Cause unix/tcl.m4 SC_PATH_X hardcoded dirs list; the rpath comes from the found XLIBSW dir
        • Prototype none (can't test; OpenWindows is obsolete, and Solaris 11 uses /usr/lib/64 X libs via pkg-config/xmkmf)
        • Suggested ticket comment OpenWindows is gone from current Solaris releases, and nobody can test this. Workaround: pass --x-libraries=/usr/openwin/lib/64. Suggest closing as out of date.
        💬1487701fff2006embedding on OS X Aqua is problematicCANNOT-TEST (macOS)

        Open ticket 1487701fff · 74. Application Embedding · filed 2006-05-12

        💬 Since triage (updated 2026-11-11)

        Status now: Open (Accepted)

        • 💬 2026-09-25 serhiy.storchaka: The console part is fixed: tcl_interactive set before TkpInit is respected. The first part is still present in the Cocoa code: TkpInit stores the first interpreter in _eventInterp ([NSApp _setup:]), all Apple event and ::tk::mac:: callbacks run in it, and nothing clears it when that interpreter is deleted. A fix could clear it with Tcl_CallWhenDeleted and let the next interpreter initialized by TkpInit take over. …
        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none. hobbs committed the tcl_interactive part (console shown only if tcl_interactive is not already set) in 2006. The "first interp only" part involves Carbon AE/linkvar code that has since been rewritten for Cocoa.
        • Cause macosx/tkMacOSXInit.c TkpInit (2006 Carbon code, since rewritten)
        • Prototype none
        • Suggested ticket comment The console part was addressed (tcl_interactive can be preset). The Carbon code has since been replaced by the Cocoa port. Needs re-checking on current macOS by someone with a Mac, otherwise close as out of date.
        ✅1481496fff2006help menu entries block the applicationCANNOT-TEST (macOS; also no pure-Tcl repro, and it depends on the reporter's C extension)

        Open ticket 1481496fff · 11. Aqua Menus · filed 2006-05-04

        ✅ Since triage (updated 2026-11-11)

        Status now: Closed (Out of Date)

        • 💬 2026-09-30 serhiy.storchaka: This was reported for the Carbon-based Aqua port of Tk 8.4, which has been replaced by the Cocoa port, and it could not be reproduced without the application's C extension. Closing as out of date. A similar symptom on the current Cocoa port, a hang when a menu command opens a modal dialog, is 33b3af1cea (macOS 27). As there, deferring the command with after idle avoids it.
        Full verdict

        CANNOT-TEST (macOS; also no pure-Tcl repro, and it depends on the reporter's C extension)

        • Platform Aqua
        • Tested none
        • Cause unknown (Carbon-era help menu handling, since replaced by Cocoa)
        • Prototype none
        • Suggested ticket comment No reproducer, and the Carbon menu code it concerns has been replaced by the Cocoa port. Closing as out of date. Please reopen with a script if it still happens with Tk 8.6/9.x.
        ⏳1432915fff2006Widget borders in canvas wrong with OS XCANNOT-TEST (macOS; Carbon-era drawing, replaced by the Cocoa port)

        Open ticket 1432915fff · 02. Appearance · filed 2006-02-16

        ⏳ Since triage (updated 2026-11-11)

        Status now: Pending

        • 💬 2026-09-25 serhiy.storchaka: The Carbon drawing code shown here has been replaced since the Cocoa port, and several redraw problems with windows embedded in a canvas were fixed recently (e.g. 61cb40b01b). Can someone on macOS check whether moving an optionmenu and a small button in a canvas still leaves artifacts with 8.6.18 or 9.0?
        • 💬 2026-09-25 kevin_walzer: I need a sample script, not just a screen shot, to test this.
        Full verdict

        CANNOT-TEST (macOS; Carbon-era drawing, replaced by the Cocoa port)

        • Platform Aqua
        • Tested none
        • Cause unknown (2006 Carbon drawing of embedded widgets)
        • Prototype none
        • Suggested ticket comment This is about the old Carbon drawing code, which has been replaced by the Cocoa port. Please re-test with Tk 8.6/9.x on current macOS, otherwise close as out of date.
        ⬜1374556fff2005sparc solaris 8 test results in lots of errorsCANNOT-TEST (Solaris 8/SPARC + Exceed, Tk 8.4.12 RC; we have no such platform, and the attached logs are from 2005)

        Open ticket 1374556fff · None · filed 2005-12-06

        Full verdict

        CANNOT-TEST (Solaris 8/SPARC + Exceed, Tk 8.4.12 RC; we have no such platform, and the attached logs are from 2005)

        • Platform build
        • Tested none
        • Cause unknown; test failures on an obsolete platform/X server
        • Prototype none
        • Suggested ticket comment Obsolete platform and version (8.4.12 RC). Closing; please open a new ticket if the test suite still fails on a current Solaris/illumos with 8.6 or 9.x.
        💬1244178fff2005TkAqua: command-n doesn\'t work when minimizedCANNOT-TEST (macOS)

        Open ticket 1244178fff · 11. Aqua Menus · filed 2005-07-25

        💬 Since triage (updated 2026-11-11)

        Status now: Open

        • 💬 2026-09-28 serhiy.storchaka: This was reported for the Carbon-based TkAqua, which was replaced with Cocoa in 8.5.7/8.6, where the menu key equivalents are handled by NSMenu. The binding bind . cannot work when "." is withdrawn or minimized, since then it has no focus, but the menu accelerator should. Does this still happen with 8.6 or 9.x?
        • 💬 2026-09-28 kevin_walzer: No, still does not work on tip of 9.1.. Some bindings are installed by default, such as Command-Q, but custom ones like this require a window to have focus.
        Full verdict

        CANNOT-TEST (macOS)

        • Platform Aqua
        • Tested none
        • Cause unknown. With . withdrawn or iconified there is no key window, so no Tk window gets the key event and only the menu key equivalent flashes. The menu entry's -command should still fire via the NSMenu key equivalent.
        • Prototype none
        • Suggested ticket comment Needs retesting on current macOS with Tk 8.6/9.0 (8.4.9-era report).
        ⏳1233014fff2005Alt / Command mismatchCANNOT-TEST (macOS). Ticket is Pending (Out of Date) after serhiy's 2026-09-23 comment.

        Open ticket 1233014fff · 01. Bindings · filed 2005-07-05

        ⏳ Since triage (updated 2026-11-11)

        Status now: Pending (Out of Date)

          Full verdict

          CANNOT-TEST (macOS). Ticket is Pending (Out of Date) after serhiy's 2026-09-23 comment.

          • Platform Aqua
          • Tested none. Reading the code: Command gives Meta_L/R and Mod1, Option gives Alt_L/R and Mod2, which matches doc/bind.n.
          • Cause none known on main
          • Prototype none (macOS only)
          • Suggested ticket comment No new information since the Pending status. Close as Out of Date unless a macOS user can still reproduce it.
          💬1155596fff2005TkAqua: display issues for buttons with image with alphaCANNOT-TEST (macOS); probably obsolete

          Open ticket 1155596fff · 66. Aqua Window Operations · filed 2005-03-03

          💬 Since triage (updated 2026-11-11)

          Status now: Open (Accepted)

          • 💬 2026-09-25 serhiy.storchaka: The QuickDraw code is gone since the Cocoa port, and since 8.6.12 photo images on macOS are drawn with TkpPutRGBAImage, which composites the alpha channel directly, without reading back the background with XGetImage. So a button with a partially transparent image should now be blended over the native button. Can someone on macOS confirm with 8.6.18 or 9.0?
          • 💬 2026-09-25 kevin_walzer: Can't test without an image that shows the problem - the images listed as attachments are not included - but this is a 20-year-old bug submitted against an obsolete implementation. I think it's safe to close this.
          Full verdict

          CANNOT-TEST (macOS); probably obsolete

          • Platform Aqua
          • Tested X11 main: button, menubutton and label with a partially transparent photo display without errors.
          • Cause the report is about Carbon/QuickDraw code (OpenCPicture in TkMacOSXDrawControl), which was removed with the Cocoa port (8.6).
          • Prototype none
          • Suggested ticket comment The QuickDraw/OpenCPicture code path no longer exists since the Cocoa port. Please reopen if a button with an alpha image still draws wrongly on current macOS.
          ⏳1121156fff2005requirement for 'lib64' checksCANNOT-TEST (old /usr/X11R6 multilib layout, SuSE 9.2); not an issue on current Linux

          Open ticket 1121156fff · 84. Unix Build · filed 2005-02-12

          ⏳ Since triage (updated 2026-11-11)

          Status now: Pending (Out of Date)

          • 💬 2026-09-25 serhiy.storchaka: On current systems libX11 is in a directory the linker searches by default, so AC_PATH_X succeeds and the fallback directory list in SC_PATH_X (which has no lib64 directories) is not used. With an old /usr/X11R6 layout, --x-libraries=/usr/X11R6/lib64 works around it.
          Full verdict

          CANNOT-TEST (old /usr/X11R6 multilib layout, SuSE 9.2); not an issue on current Linux

          • Platform build
          • Tested configure on current Linux: AC_PATH_X finds X in the default paths ("libraries , headers") and the build links fine. The fallback lists in unix/tcl.m4:SC_PATH_X still have no lib64 directories.
          • Cause the SC_PATH_X fallback directory list (used only when AC_PATH_X fails) only has 32-bit lib directories. Modern distributions keep X11 in /usr/lib64 or /usr/lib/<triplet>, which the linker searches by default, so the fallback is not needed.
          • Prototype none (it would only be adding /usr/X11R6/lib64 etc. to the fallback list, and tcl.m4 must stay in sync with Tcl). Side note: the same configure run also prints a spurious "checking for X11 header files... couldn't find any!"; I have not investigated it.
          • Suggested ticket comment Current autoconf AC_PATH_X finds X11 in the standard multiarch/lib64 locations, so the fallback in SC_PATH_X is not used on modern systems. Closing as outdated unless someone still hits this.
          ✅1080646fff2004test suite failuresCANNOT-TEST (Solaris/8.4.9). Most of the listed failures do not occur on main. Only focus-5.1 still fails, with the same result.

          Open ticket 1080646fff · 86. Test Tools · filed 2004-12-07

          ✅ Since triage (updated 2026-11-11)

          Status now: Closed (Out of Date)

          • 💬 2026-09-28 serhiy.storchaka: With 9.1 on Linux, all these tests pass with fvwm, except font-45.4, which no longer exists. unixWm-50.1 fails without a window manager, as expected. Only unixWm-4.3 still fails with metacity (GNOME 2 window manager): a toplevel moved to +20+0 while withdrawn is shown at +-6+-23 after deiconifying. unixWm-4.1 and unixWm-4.2 (the same with y 80 and 40) pass. This looks specific to that window manager.
          Full verdict

          CANNOT-TEST (Solaris/8.4.9). Most of the listed failures do not occur on main. Only focus-5.1 still fails, with the same result.

          • Platform build/test (Solaris 8, 8.4.9)
          • Tested main (wt build) under Xvfb with openbox and with no WM. Passing: font.test, frame.test (all, including -use tests 3.9/3.10), unixWm.test (4.3, 48.13 and 50.x pass. The 59.x failures are only caused by the at-spi warning on stderr). focus-5.1 fails with .t .t {} (expected .t {} {}), exactly as reported. focus-2.6/2.7/2.9 also fail under Xvfb.
          • Cause focus-5.1/2.x depend on the focus behaviour of the X server/WM (the send background app taking focus). This looks environmental, not a Tk bug.
          • Prototype none
          • Suggested ticket comment This is an 8.4.9/Solaris test-run report. On current main only focus-5.1 (and 2.6/2.7/2.9) still fail, and only under Xvfb, so they depend on the environment. Suggest closing and opening a specific ticket for the focus tests if needed.
          ⬜1032953fff2004menubutton: can't set fg on aqua Tcl/TkCANNOT-TEST (macOS)

          Open ticket 1032953fff · 11. Aqua Menus · filed 2004-09-22

          Full verdict

          CANNOT-TEST (macOS)

          • Platform Aqua
          • Tested none. Per the 2019 comment, -foreground and -activeforeground work on 8.6.9, while -background only colours the area around the button and -activebackground has no effect.
          • Cause native Aqua menubutton rendering
          • Prototype none
          • Suggested ticket comment (for a macOS tester) The -foreground part appears fixed since 8.6.9. The remaining -background/-activebackground limitation should either be documented in menubutton.n or the ticket closed.
          ⬜1032943fff2004menubutton indicator always on for aqua tkCANNOT-TEST (macOS)

          Open ticket 1032943fff · 11. Aqua Menus · filed 2004-09-22

          Full verdict

          CANNOT-TEST (macOS)

          • Platform Aqua
          • Tested none
          • Cause native Aqua menubutton draws a popup arrow regardless of -indicatoron (unverified with the current macosx/tkMacOSXMenubutton.c)
          • Prototype none
          • Suggested ticket comment Needs re-checking on current macOS with 8.6/9.x. The Aqua menubutton implementation has been rewritten since 8.4.
          ⏳1032907fff2004menubutton fg color changes unexpectedly under Aqua TkCANNOT-TEST (macOS)

          Open ticket 1032907fff · 11. Aqua Menus · filed 2004-09-22

          ⏳ Since triage (updated 2026-11-11)

          Status now: Pending

          • 💬 2026-09-25 serhiy.storchaka: The Aqua menubutton drawing code has been replaced since 8.4: the text is now drawn with the menubutton's own -foreground/-activeforeground GCs over a native popup button. Can someone on macOS check whether the script in the description still shows the problem with 8.6.18 or 9.0?
          • 💬 2026-09-25 kevin_walzer: Issue not present on tip of Tk 9.1.
          Full verdict

          CANNOT-TEST (macOS)

          • Platform Aqua
          • Tested none (the reporter states X11 is not affected)
          • Cause in the old 8.4 Aqua drawing code, colour state leaked between a bitmap button and a menubutton. That code was replaced in 8.5/8.6, so this is probably obsolete.
          • Prototype none
          • Suggested ticket comment Probably obsolete, since the 8.4 Aqua button/menubutton drawing code no longer exists. Can someone with macOS confirm it with 8.6/9.x, so the ticket can be closed?
          ⏳1012854fff2004photo image write appears to fail under TclTkAquaCANNOT-TEST (macOS; also depends on the Img extension's "window" format, not Tk)

          Open ticket 1012854fff · 66. Aqua Window Operations · filed 2004-08-20

          ⏳ Since triage (updated 2026-11-11)

          Status now: Pending

          • 💬 2026-09-25 serhiy.storchaka: As noted below, the problem is grabbing the window contents, not writing the photo. On macOS XGetImage for a window now copies the view's contents (cacheDisplayInRect in 8.6, the CALayer bitmap in 9.0), so Img's -format window should get the drawn canvas. Can someone on macOS check the script with 8.6.18 or 9.0 and a current Img?
          • 💬 2026-09-25 obermeier: Ticket can be closed. Img option "-format window" works on macOS using current Img version and Tk 8.6, 9.0 and 9.1. The above code is part of the visualtests suite of Img now.
          Full verdict

          CANNOT-TEST (macOS; also depends on the Img extension's "window" format, not Tk)

          • Platform Aqua
          • Tested none
          • Cause as hobbs noted, the failing part is not photo writing. The Img "window" format grabs window contents with XGetImage, which on OS X only works under strict viewability conditions and returns black otherwise.
          • Prototype none
          • Suggested ticket comment This is not a Tk photo write problem. The black image comes from Img's "-format window" screen grab on macOS. If it still happens with current Tk/Img it belongs to the Img tracker. Closing here.
          ⏳931324ffff2004Bwidget&TKAqua&Tabs--Mistake betwen tabsCANNOT-TEST (macOS). The reporter says Linux and Windows are fine. It is a 2004 TkAqua redraw issue, and the Aqua drawing code has been rewritten s…

          Open ticket 931324ffff · 66. Aqua Window Operations · filed 2004-04-07

          ⏳ Since triage (updated 2026-11-11)

          Status now: Pending

          • 💬 2026-09-25 serhiy.storchaka: The Aqua drawing code has been rewritten since 8.4 (Cocoa in 8.6, drawing in drawRect since 8.6.10), and several redraw problems with embedded windows in canvases were fixed recently. Can someone on macOS check whether the BWidget script in the description still misbehaves with 8.6.18 or 9.0?
          • 💬 2026-09-25 kevin_walzer: Performs expected on Tk 9.1 tip - no artifacts noted.
          Full verdict

          CANNOT-TEST (macOS). The reporter says Linux and Windows are fine. It is a 2004 TkAqua redraw issue, and the Aqua drawing code has been rewritten several times since.

          • Platform Aqua
          • Tested none
          • Cause unknown (Aqua redraw of scrolled, embedded frames in an unraised notebook page)
          • Prototype none
          • Suggested ticket comment Can someone on macOS check whether this still happens with Tk 8.6.16 / 9.0 using the BWidget script from the description? Otherwise it will be closed as out of date.
          ⬜895082ffff2004TkAqua: -use option to wish not implemented on OS XCANNOT-TEST (macOS). This is effectively a won't-fix/limitation: macosx/tkMacOSXEmbed.c:Tk_UseWindow still limits containers to the same process ("…

          Open ticket 895082ffff · 75. wish · filed 2004-02-11

          Full verdict

          CANNOT-TEST (macOS). This is effectively a won't-fix/limitation: macosx/tkMacOSXEmbed.c:Tk_UseWindow still limits containers to the same process ("we do not allow embedding into windows belonging to a different process"). Cocoa gives no way to embed another process's NSView.

          • Platform Aqua
          • Tested source reading only (main)
          • Cause macosx/tkMacOSXEmbed.c:Tk_UseWindow looks the container up with Tk_IdToWindow in the current application only.
          • Prototype none (not feasible; at most a doc note in wish.1/toplevel.n that -use works on macOS only within one process)
          • Suggested ticket comment On macOS, embedding with -use works only within the same process (tkMacOSXEmbed.c), and Cocoa offers no cross-process embedding. Suggest documenting the limitation and closing.
          ✅853398ffff2003unixEmbed-5.1 EmbedFocusProc procedure, FocusIn events FAILECANNOT-TEST (Tru64 Unix); NOT-REPRODUCED on Linux

          Open ticket 853398ffff · 86. Test Tools · filed 2003-12-03

          ✅ Since triage (updated 2026-11-11)

          Status now: Closed (Fixed)

          • 🔧 2026-09-26 check-in 9764b127b8 on embed-focusin-test, in main, 9.0: Fix 853398ffff: unixEmbed-5.1 failed sporadically, because the child process returned the result before handling the FocusIn event. Process pending events fir
          • 🔧 2026-10-15 check-in cf6996f47a on trunk, in main, 9.0: Fix 853398ffff: unixEmbed-5.1 failed sporadically
          • 🔧 2026-10-15 check-in d3532aec4b on core-9-0-branch, in main, 9.0: Fix 853398ffff: unixEmbed-5.1 failed sporadically
          • 🔧 2026-10-16 check-in 141570e59b on core-8-6-branch, in main, 9.0, 8.6: Fix 853398ffff: unixEmbed-5.1 failed sporadically. Process pending events in the child before reading the result, as on main (cf6996f47a).
          • 💬 2026-09-26 serhiy.storchaka: unixEmbed-5.1 still fails sporadically (about 1 of 6 runs on Linux): the child process returns $x before it has handled the FocusIn event. Proposed fix in branch embed-focusin-test: process pending events in the child first (update; set x), as unixEmbed-5.3 does.
          • 💬 2026-10-16 jan.nijtmans: Fixed in cf6996f47a and core-9-0-branch Many thanks!
          • 💬 2026-10-16 serhiy.storchaka: Thank you, Jan, for the review and merging. Backported to core-8-6-branch in 141570e59b: unixEmbed-5.1 failed there in 1 of 18 runs too, and passes 30 of 30 with the fix.
          Full verdict

          CANNOT-TEST (Tru64 Unix); NOT-REPRODUCED on Linux

          • Platform X11 (Tru64)
          • Tested main, unixEmbed.test on Xvfb with no WM. unixEmbed-5.1 and 5.1a pass. Side note: unixEmbed-9.2 is flaky there (failed 2 of 4 runs), which has nothing to do with this ticket.
          • Cause unknown. Most likely a timing or focus issue with the X server of that platform in 2003.
          • Prototype none (not reproducible)
          • Suggested ticket comment unixEmbed-5.1 passes on current Linux/X11 (main). The report is from Tru64 with 8.4.5, a platform that is no longer available. Closing as out of date.
          💬592096ffff2002Tk requires same build setup as TclCANNOT-TEST (Solaris, Sun cc vs gcc; only gcc available here, no clang)

          Open ticket 592096ffff · 84. Unix Build · filed 2002-08-07

          💬 Since triage (updated 2026-11-11)

          Status now: Open (Remind)

          • 💬 2026-09-25 serhiy.storchaka: Since 8.5.0 (06ad11b82c) Tk's configure computes its own search flags from its own compiler instead of copying TCL_LD_SEARCH_FLAGS/TCL_CC_SEARCH_FLAGS from tclConfig.sh, so a shared Tk built with another compiler no longer gets Tcl's -Wl,-R,... flags. Linking a static Tcl built with a different compiler still cannot work, as explained below.
          Full verdict

          CANNOT-TEST (Solaris, Sun cc vs gcc; only gcc available here, no clang)

          • Platform build
          • Tested code reading only. Tk's unix/configure.ac runs its own SC_CONFIG_CFLAGS, so compiler/linker flags come from Tk's configure, not tclConfig.sh. TCL_LD_SEARCH_FLAGS is used only at configure.ac:285.
          • Cause n/a
          • Prototype none
          • Suggested ticket comment Tk's configure computes its own compiler/linker flags (SC_CONFIG_CFLAGS) these days. We can't test Sun cc on Solaris; closing as Out of Date unless it can be reproduced with a current Tk.