"Jarrad Hope" <[EMAIL PROTECTED]> (on Fri, 16 Jun 2006 16:03:11 +0800):

  >                                     Binding Cairo
  > 
  > Straight binding since its written in C - Glitz development is slow and  
  > apparently incomplete?

i'm not so sure about glitz, but i expect it to at least get good backing 
someday, once GTK switched to cairo completely, and XGL (or that other X-GL 
binding from redhat or whatnot) is more prevalent. i mean- the whole linux/etc 
desktop will depend on it, so there's lots of interest :) i didnt really look 
into that though, this is just guesswork from the little i follow...


  > AGG is pure software rendering library, with no HW acceleration. It means  
  > that you can't use it for heavy duty realtime graphics. But still, in  
  > practice the performance is good enough for, say, casual games.
  > 
  > > Could you help us with this? Or know a way to make this easy? or any  
  > > other suggestion?
  > 
  > You can surely use AGG from pure C, for example, Carl Sassenrath with his  
  > REBOL language does namely that.
  > 
  > The first thing to do is clearly define the interface. Then, I can help  
  > you with writting a wrapper over AGG with pure C functions.
  > 
  > The interface can be something like OpenVG, for example, or, Java2D,  
  > simplified and modified for pure C specific.

interface, well- how about cairo? there are two options i could see that could 
give us best of both; both of them could be vaguely described as "having agg as 
a backend for cairo".

a) cairo-agg- antigrain as a "proper" backend for cairo, using cairo's backend 
API. cario currently sports a software-renderer, glitz and postscript. if agg 
is better than the cairo softrenderer, this would make sense.

b) cario-API wrapper on top of agg- antigrain not as a proper backend for 
cairo, but a thin lib on top of agg that implements the cairo api. instead of 
seeing cairo as a "lib proper" here, we "just" take it's C interface definition.

i have to think about both (and other options) more. meanwhile, i've started 
trying to bind cairo. it pushes my little nekobind above its limits, but we'll 
see..

as for xinf, i will likely continue using the GL api directly, if only to keep 
the possibility of doing "heavy duty realtime graphics". otoh, cairo 
implements/defines so much of what i have to do myself that i should maybe 
reconsider. ahh- so many options! :)


thanks for doing the research, anyhow. the comments from McSeem are 
interesting, his offer to help bind agg once we've defined an interface is very 
nice, too.

-dan

-- 
http://0xDF.com/
http://iterative.org/

-- 
Neko : One VM to run them all
(http://nekovm.org)

Reply via email to