>> >> As the number of drivers is quite small it won't actually speed
>> >> up FreeType. But having an initial guess for checking drivers
>> >> can certainly remove some constant time complexity required to
>> >> check through different drivers..
>>
>> > Yep.
>
> Why hasn't this been included in FreeType uptil now?
Optimizing FreeType for speed hasn't shown this as a hot spot...
On the other hand, there wasn't much effort into this direction; I can
imagine that profiling experts find a lot of places where FreeType can
be made faster. However, there must be a balance between readable
code and optimizations, and please always remember that for efficient
use FreeType output (i.e., outlines, metrics, etc.) must be cached as
much as possible.
Werner
_______________________________________________
Freetype-devel mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/freetype-devel
