[Fontconfig] Re: Strange behaviour of fc-match of fontconfig current cvs (2.3.93).

[Patrick Lam]

Quoting Zhe Su <james.su at gmail.com>:

> Hi,
>  Reverting FcFontSetMatch()  to the one in 2.3.91 does fix this
> issue. So there is something wrong in the new algorithm.
>  I think at least we should reuse FcCompare() in FcFontSetMatch()
> rather than using another piece of code. FcFontSetSort should have the
> same result with FcFontSetMatch () at this point.

Inlining and doing optimizations on FcCompare results in a huge perf gain
whenever this codepath is used.  Of course it should work correctly, but I'd
rather fix the bug than reuse FcCompare.

pat