Ryan C. Gordon
Ryan C. Gordon
This has always driven me nuts (in CMake as well as autoconf). Before we preseeded Emscripten, these checks would take an eternity. It's all compile/link time. I wonder if there...
But the problem is nothing is _wrong_ either way; we're deciding if we need our internal fallbacks for various functions or not, not whether the library will fail to build.
I'm gonna try an idea out, stay tuned.
Okay, I ran with @maia-s's idea, but "fine-grained" isn't "string functions" but "ANSI C", "POSIX", "wcs* functions" and "Other Stuff" at the moment. Basically it tries to compile them all...
(Also it just pretends the functions are void fn(void); and tries to call them. We don't care, we just want to see if the linker will resolve the symbol, but...
The linker is just looking for a symbol named "strcmp" and doesn't know anything else about it, so it works. (In C++, function information is encoded in the symbol the...
Going to try to reproduce on an older Mac, to see if this is something that was changed in macOS 15 or whatever.
Oh, interesting. I'm going to close #14627 as a dupe and deal with this here.
So I think checking for the focused window is wrong here in any case; EmscriptenFullscreenChangeEvent can tell us the element that was entering/existing fullscreen, so we can probably find the...
Okay, this _should_ be fixed now, please try the latest from revision control and let me know if it's still having problems!