Hi,
This patch is putting me in a bind: it introduces a failure which is in fact a fix. But only in Wine and in the Japanese locale!
https://test.winehq.org/data/patterns.html#riched20:txtsrv
debiant2-win32-ja-JP: txtsrv.c:768: Test succeeded inside todo block: got wrong width: 72, expected: 72 {+1}.
As far as I can tell the test is not wrong: * It succeeds in all the locales on Windows. * The todo is still needed in all the other locales in Wine: the returned width systematically falls 2 to 3 pixels short.
So I don't know how to avoid getting this failure in every nightly WineTest run.
Is it a font problem? Making the todo conditional on the locale feels iffy. Same for skipping the test based on the locale. Fix the implementation some more so it works on all locales?