[Cairo] Review of text API in new virtualized font interface

Carl Worth cworth at east.isi.edu
Tue Oct 21 12:46:36 PDT 2003


On Oct 21, Keith Packard wrote:
 > If you're going to use CSS2, you might as well stick all of the values 
 > into the typedef; it's 'free'; the underlying code can simply compare <= 
 > NORMAL and > NORMAL for not-bold/bold.

The rest of the CSS2 weights are just numbers, right? (except bolder
and lighter of course which we won't need). Perhaps we should just
ignore CSS2 altogether. One fear I have about basing the "toy" API on
CSS2 at all is that a user might feel the need to wade through that
giant spec. just to use this. I certainly wouldn't want that.

 > Style includes many attributes not covered here, this really is
 > just slant.

Since I want to ignore CSS2 now, I actually prefer slant as well. But
I'm curious, from what I read[*] CSS2 "font-style" has exactly the
values "normal", "italic", and "oblique", (and "inherit" which isn't
interesting). Did I miss something about CSS2 or did you just mean in
general that "style" suggests more than just "slant".

 > cairo_select_font/cairo_set_font should always be followed by 
 > cairo_scale_font/cairo_transform_font.

Yes. And in that case, we can just drop the cairo_font_scale and
cairo_font_transform calls too. I like that idea.

 > Freeze/thaw?  I don't see the point here.  With cairo_set_font, there may 
 > be device-specific 'issues' with scaling fonts, but that's not our problem 
 > in this part of the API.

graydon and I were chatting and perceived a potential future problem
that this could address. We'll see if it comes up or not.

 > > The freetype-specific portion of the API should be moved into a new
 > > file cairo-ft.h.
 > 
 > I thought we'd discussed this and decided that cairo.h should be 
 > sufficient on every platform and that applications can do feature 
 > dectection to discover what cairo APIs were available.

You and I had discussed this, but we didn't have a decision here on
the list yet. I was still a bit concerned about namespace pollution,
but you've now proposed a nice solution for that.

Long live the single cairo.h file.

-Carl

[*] http://www.w3.org/TR/REC-CSS2/fonts.html




More information about the cairo mailing list