Oleg Oshmyan

If I understand correctly, neither pthreads nor stdatomic.h are used `ifdef _WIN32`? If so, maybe configure should avoid running the corresponding checks (and setting compiler flags) on Windows.

Thanks! (And congrats!) I don’t mind the `TextInfo` thing being separate/later, especially seeing as fixing the whole chain of commits would’ve been a lot of work for barely any gain....

This is trivial to implement, but it will affect output bitmap sizes, which some applications currently use for extent measurement.

https://issuetracker.google.com/issues/111801338 was eventually filed with Android. It was marked as “fixed” on 2 Nov 2018, but no explanation has ever been provided.

Finally found something official: [“What’s new for text in Android Q”](https://android-developers.googleblog.com/2019/07/whats-new-for-text-in-android-q.html) [NDK reference: Font](https://developer.android.com/ndk/reference/group/font.html#asystemfontiterator_close) shows three functions that are presumably new in Android Q: ```c ASystemFontIterator *_Nullable ASystemFontIterator_open(); AFont *_Nullable...

Ah, right. 😅 We should be able to compile once and run on both new and old Androids. I’m assuming Android probably provides some way to figure out its version...

Probably related is #305, where Fansub Block doesn’t seem to get faux-bolded, much like Bona Nova in your sample. No idea whether its metrics change though.

> This is likely a regression in [ac75f67](https://github.com/libass/libass/commit/ac75f67697845efc1b0a15693c96dcc9e0f71518); we need to check _both_ conditions. Should we sneak a fix for this in 0.15.0?

`ass_fonts_update` and the `update` argument to `ass_set_fonts` are now no-ops (since the introduction of fontselect), which simplifies and unbreaks the API somewhat. `ass_set_fonts` still needs to be called after `ass_add_font`,...

For reference, what format/source are those subtitles, e. g. SRT or ASS? --- Translation for developers: apparently mpv has an option that sets the style’s `Spacing` (`\fsp`), which, of course,...