Christopher Degawa

`fcntl` doesn't exist on windows, so I assume that's where an issue would arise from. You might want to check to see if there's some way around that or see...

Tested with pkgconf's git master as well and still same behavior, so it is still present there. Bisected to the first (non-build broken) commit to https://github.com/pkgconf/pkgconf/commit/197fcadd4c44504808a44fb257275684b5207b10 Something within `pkgconf_queue_collect_dependents()` is...

Interestingly, this does not reproduce on Linux: ```bash [ccom1@ccom-pc pkgconf (master 0226cdd) 15:42:03]$ ./bui/pkgconf --static --libs opusurl --version 1.9.3 [ccom1@ccom-pc pkgconf (master 0226cdd) 15:42:07]$ ./bui/pkgconf --static --libs opusurl -lopusurl -lopusfile...

nvm, making qsort match the sorting doesn't change anything wrt the output, wrong tree bark

Wait, it breaks the linux output to match the Windows output, huh

```diff diff --git a/libpkgconf/queue.c b/libpkgconf/queue.c index 61637e3..53f6241 100644 --- a/libpkgconf/queue.c +++ b/libpkgconf/queue.c @@ -144,7 +144,9 @@ dep_sort_cmp(const void *a, const void *b) const pkgconf_dependency_t *depA = *(void **) a; const...

Of course, the above relies on stuff that I'm not sure if defined for qsort and pointers, but 🤷

opened an upstream issue https://github.com/pkgconf/pkgconf/issues/268

I think it's more like the current mingw-w64-i686 packages will be frozen at their current state and none of those packages will receive any more updates, maybe not even bugfixes...

Ahh, I see, nevermind then