GPII-3759: Bodge quality fix to allow language solution to have its settings applied last

[amb26]

No description provided.

[gpii-bot]

CI job passed: https://ci.gpii.net/job/universal-tests/1691/

[javihernandez]

@amb26, @stegru, @JavierJF, @cindyli what do we want to do with this one?

Remember that this PR was created without having a clear idea of the reasons behind the problems we were experiencing when changing highContrast and language at the same time.

Later on, we found out that some of the problems were (among others already solved):

1. The background wallpaper wasn't being restored - see GPII-3844: Prevent system wallpaper corruption when setting high-contrast mode #772
2. We need to know the localised name of the high contrast theme in any of the languages that are installed on the machine - see https://issues.gpii.net/browse/GPII-3845 and GPII-3845: Fix for wrong localization retrieval for high-contrast theme name id javihernandez/windows#14

Shall we mothball this for now until we're done properly fixing these two above and then decide if we need this improvement in place right away (or it can wait) ?

[amb26]

Happy to mothball this if it isn't effective - have we ever tried building an installer without it and seeing if the quality of fix gets any worse?

[javihernandez]

Happy to mothball this if it isn't effective - have we ever tried building an installer without it and seeing if the quality of fix gets any worse?

Nope, we haven't created such installer. In any case, after having gone over the reports again, the problem could also be reproduced when changing the DPI screen scaling (in addition to the language one). In fact, we couldn't reproduce the problem ourselves, we just observed the problem happening in testers computers during a live testing session.

What we know for sure is that the language switch restarts the explorer.exe, which is a bit "too much" for both Windows (which we assume that certain settings change might be applied while explorer.exe is not running) and us (since we trigger the application of the settings asynchronously and there's no control over the application of them).

I do know that this is something that both of us would like to see mothballed, but I'd say that we should prevent ourselves against this kind of "randomness" when applying settings and when a language switch is performed.

@amb26, @stegru, @JavierJF, @cindyli, what are your thoughts on this?

[amb26]

Looks like mothball to me!