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

[PATCH 0/1] MR10962: winewayland.drv：Generate a unique winewayland tag for each Wine Bottle
by pan hui (＠panhui) May 20, 2026

May 20, 2026

Currently, WINEWAYLAND_TAG_MIME_TYPE is a fixed string "application/x.winewayland.tag", all wine bottles share the same tag, and copy-paste operations between different bottles are ignored. Generate a unique winewayland tag for each wine Bottle, and only the tag issued by itself are ignored. The tags of different bottles are different, so there is no misjudgment. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10962

2 1

[PATCH v2 0/2] MR10960: scrrun/dictionary: Match Empty keys against zero-valued and empty-string keys.
by Francis De Brabandere (＠francisdb) May 20, 2026

May 20, 2026

An uninitialized VBScript variable is Empty, so scripts can end up using Empty as a dictionary key alongside a numeric 0 or "". Native treats those as the same key; Wine kept Empty distinct from every other type. Match an Empty key against any numeric zero or the empty string, while keeping it distinct from Null and from non-zero/non-empty keys. -- v2: scrrun/dictionary: Match Empty keys against zero-valued and empty-string keys. scrrun/tests: Add tests for Empty dictionary keys. https://gitlab.winehq.org/wine/wine/-/merge_requests/10960

3 4

[PATCH 0/2] MR10960: scrrun/dictionary: Match native key semantics for Boolean, Currency, and Empty keys
by Francis De Brabandere (＠francisdb) May 20, 2026

May 20, 2026

Boolean and Currency keys were rejected outright; hash them by their numeric value like the other numeric types. Compare numeric keys at double precision instead of single, so that values differing only beyond float precision (e.g. 16777216 and 16777217) remain distinct keys. Treat Empty as the zero/default value of the other key's type, matching any numeric zero (including False and date 0) or the empty string. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10960

3 4

Re: [PATCH v3 0/6] MR10938: msxml3: Reimplement XPath support without libxml2.
by Bernhard Übelacker (＠bernhardu) May 20, 2026

May 20, 2026

@nsivov, I received in this ASan runs some reports from msxml3:domdoc: https://gitlab.winehq.org/bernhardu/wine/-/jobs/268764 https://gitlab.winehq.org/bernhardu/wine/-/jobs/268763 I have not yet looked into it in more detail, but I think they may relate to this merge request. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10938#note_140764

1 0

[PATCH v5 0/6] MR10716: dsound: Speed up resampling, part 6
by Anton Baskanov (＠baskanov) May 20, 2026

May 20, 2026

-- v5: dsound: Add an SSE version of downsample. dsound: Add an SSE version of upsample. https://gitlab.winehq.org/wine/wine/-/merge_requests/10716

4 17

Re: [PATCH v2 0/1] MR10948: include: Update CERT_KEY_CONTEXT.
by Hans Leidekker (＠hans) May 20, 2026

May 20, 2026

On Wed May 20 04:09:13 2026 +0000, Trent Waddington wrote: > We were recently investigating this code when running `dotnet restore` > and it would appear that crypt32 on Windows doesn't do this dereference. > The value of pvData is stored into the parameter and will be retrieved > later. Doing the dereference here results in the contents of the handle > being stored and causes a later crash. i.e., the documentation says what > you're doing here is right, but the implementation doesn't match that. > It would be great if you could confirm, as we have questioned our sanity > investigating this. I don't know if this is right, this patch only changes the code to use the appropriate union field. I looked into this in the context of this bug: https://bugs.winehq.org/show_bug.cgi?id=59757 Does 'dotnet restore' also pass a CERT_KEY_CONTEXT instead of a NCRYPT_KEY_HANDLE by any chance? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10948#note_140760

1 0

[PATCH 0/1] MR10955: Revert "dxcore: Prefer native."
by Elizabeth Figura (＠zfigura) May 20, 2026

May 20, 2026

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/10955

3 3

[PATCH 0/2] MR10949: winex11.drv: Return WithdrawnState when get_window_wm_state() fails.
by Zhiyi Zhang (＠zhiyi) May 20, 2026

May 20, 2026

KWin deletes WM_STATE property from a window when entering WithdrawnState. Please see X11Window::exportMappingState() in the KWin source code. So when get_window_wm_state() fails, -1 is returned before this patch. Since -1 is not a valid WM_STATE, it might cause Wine to expect an invalid value. We also can't simply ignore the invalid -1 because we might be expecting the WM_STATE property notify event to change to the desired state. Treating missing WM_STATE property as WithdrawnState seems like the best way to go. Help fix RiME(493200) sometimes fails to restore from Alt+Tab. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10949

3 6

[PATCH 0/2] MR10950: Draft: shell32: Add an Info icon.
by Stefan Dösinger (＠stefan) May 20, 2026

May 20, 2026

Hi Alex, I ran into an application that needed SIID_INFO. I noticed you added a lot of icons to shell32, so maybe you can help me with two issues: 1) On my WinXP and Win11, the (i) Icon has index 221, not 215. The power off icon however has index 215. Can you double check if you used the correct value for IDI_SHELL_POWER_OFF back in 9954c08dc8dc? 2) It seems the other shell icon SVGs have a rasterized image embedded, which tools/buildimage needs. For this .svg it writes "no render directive found in dlls/shell32/resources/info.svg". How did you generate those for the icons you added? Other Wikimedia Commons Tango icons don't seem to have them, so I guess they were added by one of us. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10950

3 4

[PATCH v2 0/1] MR10914: win32u: Allow OpenGL context / HDC pixel format mismatch.
by Rémi Bernon (＠rbernon) May 20, 2026

May 20, 2026

This may happen if HDC is from a Pbuffer, and context was created separately. Let's not be restrictive if host GL allows it. Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=59555 -- v2: win32u: Allow OpenGL context / HDC pixel format mismatch. https://gitlab.winehq.org/wine/wine/-/merge_requests/10914

2 2