[Bug 51478] New: dinput:keyboard breaks user32:input's test_keyboard_layout_name() in some locales ()

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=51478

            Bug ID: 51478
           Summary: dinput:keyboard breaks user32:input's
                    test_keyboard_layout_name() in some locales ()
           Product: Wine
           Version: 6.10
          Hardware: x86-64
                OS: Windows
            Status: NEW
          Severity: normal
          Priority: P2
         Component: user32
          Assignee: wine-bugs at winehq.org
          Reporter: fgouget at codeweavers.com

Created attachment 70313
  --> https://bugs.winehq.org/attachment.cgi?id=70313
Add test_keyboard_layout_name() traces

The WineTest runs show the following test_keyboard_layout_name() failure in
some locales:

https://test.winehq.org/data/patterns.html#user32:input

input.c:3169: Test failed: Could not find keyboard layout 0000000004110411 in
preload list

This is 100% reproducible in the following configurations (only the keyboard
layout id differs):
  win7_newtb-w7u-es
  win7_newtb-w7u-pt-PT
  win2004_newtb-w10pro64-ar-64
  win2004_newtb-w10pro64-he-64
  win2004_newtb-w10pro64-hi-64
  win2004_newtb-w10pro64-ja-64
  win2004_newtb-w10pro64-ko-64
  win2004_newtb-w10pro64-pt-BR-64
  win2004_newtb-w10pro64-ru-64
  win2004_newtb-w10pro64-zh-CN-64

Notes:
* Interestingly this means this failure does not happen in the English, French,
German and Greek locales.
* Also user32:input only fails on the first run that follows dinput:keyboard,
and does not ever fail again, even if dinput:keyboard is run again.

In the w10pro64_pt_BR case, adding extra traces shows that the root of the
issue is that GetKeyboardLayoutList() returns 10 instead of 13. As a result
test_keyboard_layout_name() fails to retrieve all the Keyboard Layout\\Preload
values (there is 12 of these, not 10 or 13!).

Also the items are not in the same order in layouts_preload and layouts so this
causes some entries to be missing.

More notes:
* The ok() message is wrong: it is not layout we are looking for but
layouts[i], aka klid.

* I don't know why the GetKeyboardLayoutList() return value changes in the
first run that follows dinput:keyboard, but I'm not sure it makes sense to
assume it matches the number of entries in the Preload key in the first place.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.