[Fontconfig] Application startup performance

[Raimund Steger]

On Tue, 15 Mar 2016 00:22:30 +0000
Nick Alcock <nick.alcock at oracle.com> wrote:

> On 10 Mar 2016, Behdad Esfahbod outgrape:
> 
> > Any resolution here?  Current master's behavior of stating all fonts is
> > definitely very bad.  Anyone volunteer to clean up the cache code a bit?
> 
> I submitted a patch that cleans this up a bit for unchanging fonts -- it
> was semi-pulled by Akira (I think) for later investigation. Did anything
> come of it?

Do you mean the one that removed qsort and used d_type instead of stat?

qsort is negligible especially compared to NFS I/O, it's not I/O bound. I think you would have a hard time make it even show up in a profiler.

d_type will speed up things but only if it's available (non-SysV). But I think that these Fedora bugreports might not have happened with nanosecond mtime, thus potentially rendering the whole FcCacheDirsValid execution path redundant.

Raimund


-- 
Worringer Str 31 Duesseldorf 40211 DE  home: rs at mytum.de
+49-179-2981632 icq 16845346           work: rs at interface-ag.de