April 2026 - Wine-GitLab - list.winehq.org

Re: [PATCH v2 0/4] MR10654: winegstreamer: Fix wg_parser/native video stride mismatch.
by Yuxuan Shui (＠yshui) April 29, 2026

April 29, 2026

On Wed Apr 29 13:59:58 2026 +0000, Yuxuan Shui wrote: > OK I tried rawvideo in avi container. and additionally i think I must > avoid doing pixel format conversion to avoid involving mf transforms? > after doing that, i noticed that the `IMF2DBuffer` interface was gone > from the samples. so likely that was mf transforms related. > now what I saw was quite odd, my input video is i420 161x120, the output > i got is somehow 160x121. huh? > the output buffer size is 29040, which is `160x121 * 3/2`. but > `MF_MT_FRAME_SIZE` still reports 161x120. > (not sure if this information is relevant, but i also tried using a raw > rgb32 video with pixel format conversion to yuv, which i believe would > be using transforms. this time I got buffer size 29160, which is > `162x120 * 3/2`, which is exactly what's expected. and `IMF2DBuffer` > interface again becomes available.) can we maybe conclude odd width video is bugged on Windows? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10654#note_138198

1 0

Re: [PATCH v2 0/4] MR10654: winegstreamer: Fix wg_parser/native video stride mismatch.
by Yuxuan Shui (＠yshui) April 29, 2026

April 29, 2026

On Mon Apr 20 21:30:57 2026 +0000, Yuxuan Shui wrote: > Oh right, I am dumb. I was only thinking about wine and forgot h264 > would also be going through IMFTransform on Windows. I'll do what you suggested. OK I tried rawvideo in avi container. and additionally i think I must avoid doing pixel format conversion to avoid involving mf transforms? after doing that, i noticed that the `IMF2DBuffer` interface was gone from the samples. so likely that was mf transforms related. now what I saw was quite odd, my input video is i420 161x120, the output i got is somehow 160x121. huh? the output buffer size is 29040, which is `160x121 * 3/2`. but `MF_MT_FRAME_SIZE` still reports 161x120. (not sure if this information is relevant, but i also tried using a raw rgb32 video with pixel format conversion to yuv, which i believe would be using transforms. this time I got buffer size 29160, which is `162x120 * 3/2`, which is exactly what's expected. and `IMF2DBuffer` interface again becomes available.) -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10654#note_138197

1 0

[PATCH v5 0/1] MR10725: win32u: Set the normal window position without going through NtUserSetInternalWindowPos().
by Zhiyi Zhang (＠zhiyi) April 29, 2026

April 29, 2026

Fix a regression from be60f779, the intent of which was to restore the win32 window position to the current host window position. However, NtUserSetInternalWindowPos() eventually calls NtUserSetWindowPos() and NtUserShowWindow(). We don't want that because they might generate window position change messages that applications might not expect. Fix Homeworld Remastered Collection (244160) Homeworld 2 Classic black screen after restoring from the minimized state in fullscreen mode due to an unexpected window size. -- v5: win32u: Set the normal window position without going through NtUserSetInternalWindowPos(). https://gitlab.winehq.org/wine/wine/-/merge_requests/10725

3 3

[PATCH v6 0/2] MR10340: comctl32/tooltips: suppress icon and title when title is NULL or empty.
by Larry Starr (＠starrl) April 29, 2026

April 29, 2026

Per MSDN, TTM_SETTITLE with a NULL or empty title string should suppress both the title row and any associated icon. Previously an empty string was treated the same as a non-empty title, causing the icon to still be displayed and the tooltip to be incorrectly sized. Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=58236 Signed-off-by: Larry Starr <Larry_A_Starr(a)yahoo.com> -- v6: comctl32/tooltips: suppress icon and title when title is NULL or empty. comctl32/tooltips/tests: add regression test for TTM_SETTITLE with NULL or empty title. https://gitlab.winehq.org/wine/wine/-/merge_requests/10340

3 5

[PATCH 0/1] MR10676: gdi32/uniscribe: Correct the script tag for Tamil.
by समीरसिंह Sameer Singh (＠ss141309) April 29, 2026

April 29, 2026

The script tag for the Tamil script is 'tml2' and not 'tam2'. The fix turned out to be easier than expected. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10676

4 4

[PATCH v5 0/2] MR10612: gdi32/uniscribe: Fix heap buffer overflow in ScriptStringAnalyse().
by समीरसिंह Sameer Singh (＠ss141309) April 29, 2026

April 29, 2026

Hi, I came across this bug while solving another issue. Steps to reproduce: 1. Open wine notepad 2. Start spamming any character that requires mark_invalid_combinations() in the notepad, such as: َ (0x064e Arabic Fatha) 3. At nearly 41 inserted Fatha, notepad will crash -- v5: gdi32/uniscribe: Resize the buffers if glyph count exceeds limit. gdi32/uniscribe: Perform bounds check in insert_glyph(). https://gitlab.winehq.org/wine/wine/-/merge_requests/10612

4 4

[PATCH 0/1] MR10439: crypt32: Protect certificate context from repeated releases.
by Dmitry Timoshkov (＠dmitry) April 29, 2026

April 29, 2026

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/10439

3 16

[PATCH v2 0/4] MR10775: vbscript: Fix and expand test coverage for BSTR comparison. (alternate)
by Francis De Brabandere (＠francisdb) April 29, 2026

April 29, 2026

Alternate version of !10766 This replaces runtime VT_RESERVED tracking with compile-time is_literal_expr at the AST level: comparison opcodes (equal/nequal/gt/gteq/lt/lteq/case) carry a flag arg with bit 0 = left-literal, bit 1 = right-literal, and var_cmp dispatches on those flags. No runtime stripping, no OP_strip_reserved, and Const references stay non-literal because they're EXPR_MEMBER nodes even though the value is inlined. -- v2: vbscript: Match Windows BSTR-vs-numeric and BSTR-vs-Boolean comparison. https://gitlab.winehq.org/wine/wine/-/merge_requests/10775

2 4

[PATCH 0/5] MR10776: dinput: Add proper support for DIPROP_JOYSTICKID and GUID_Joystick.
by Rémi Bernon (＠rbernon) April 29, 2026

April 29, 2026

Supersedes https://gitlab.winehq.org/wine/wine/-/merge_requests/10755 with some style fixes and simplified joystick ID allocation, moved to a dedicated function. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10776

3 5

[PATCH 0/4] MR10775: vbscript: Fix and expand test coverage for BSTR comparison. (alternate)
by Francis De Brabandere (＠francisdb) April 29, 2026

April 29, 2026

Alternate version of !10766 This replaces runtime VT_RESERVED tracking with compile-time is_literal_expr at the AST level: comparison opcodes (equal/nequal/gt/gteq/lt/lteq/case) carry a flag arg with bit 0 = left-literal, bit 1 = right-literal, and var_cmp dispatches on those flags. No runtime stripping, no OP_strip_reserved, and Const references stay non-literal because they're EXPR_MEMBER nodes even though the value is inlined. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10775

2 5