Re: [Freetype] Re: [Render] Re: OpenType support in Pango

[Owen Taylor]

Werner LEMBERG <address@hidden> writes:

> > It's true that the implementation in FT1 where the string is copied
> > over for every lookup applied is not very suited to applying large
> > sets of lookups, however I think the model:
> > 
> >  - Set per-glyph properties determining which lookups should be
> >    applied.
> >  - Apply a series of lookups to the entire string.
> > 
> > Is fine, and a lot simpler than alternatives.
> 
> Note that the FT1 code has been written mainly for demonstration
> purposes.  I'm not a specialist for string manipulation, and I'm sure
> that a more efficient implementation is possible.

I understand that, and I'm quite grateful that there is code at
all of any sort.
 
> What exactly do you mean with `alternatives'?  Data-structures
> resp. implementation issues or the lookup algorithm itself?  If you
> mean the latter, can you please elaborate?

By "alternatives" I meant here "alternative APIs" - such as having to
explicitely apply individual lookups to individual groups of
glyphs. I was just saying that I think the basic idea of the public
interface you set up is good.

I don't think much actual modification is possible or desirable
to the details of the lookup algorithm, though if shaping turns
out to be a significant bottleneck that could be examined.

If I was trying to rewrite for performance, the first thing I'd
look at would be doing most of the lookups in place instead
of copying from one string to another, but even that probably
isn't worth doing without some performance evaluation.

Regards,
                                        Owen