XShape example?

[Carsten Haitzler (The Rasterman)]

On Sat, 29 Oct 2005 17:50:11 +1000 Russell Shaw <rjshaw at netspace.net.au>
babbled:

> Matt Garman wrote:
> > On Sat, Oct 29, 2005 at 01:38:59PM +0900, Carsten Haitzler wrote:
> > 
> >>On Fri, 28 Oct 2005 23:20:17 -0500 Matt Garman
> >><garman at raw-sewage.net> babbled:
> >>
> >>>eyes looks as though it's transparent).  [BTW, I know it's not true
> >>>transparency, but as long as it *looks* transparent, that's good
> >>>enough for me.]
> >>
> >>it is true transparency. no pixels exist in the "shaped out" areas
> >>- you see right through it - events go through the holes too. its
> >>the oldes form of transparency x has - but effectively only
> >>provides 1 bit of accuracy :) (i wont go into how its implemented
> >>with rectangle lists) :)
> > 
> > 
> > Ahh, now that I have a working example, I see that it is in fact
> > true transparency.
> > 
> > When you say that this is the "oldest form of transparency x has",
> > does that imply there are *better* ways of doing transparency?
> > 
> > As a general disclaimer, I'm just getting into Xlib programming, so
> > it's safe to assume total ignorance on my part!
> 
> You should get the Xlib Programming Manual, Xlib Reference Manual,
> and others in the set if interesting. ISBN: 1-56592-002-3
> 
> It explains all about colour-planes.
> 
> With core X, to blend colours in a mathematically linear way as you'd
> expect, you'd need to read the image from the server, do your own blending
> with a separate alpha-channel image, then write it back to the server, which
> makes for very slow performance.
> 
> There's X extensions (Render extension IIRC) that accepts alpha-channel
> data directly.
> 
> I've been tempted to add alpha-channel colours to a bunch of the X core
> functions, but with slightly different names and as an extension.
> However, i can't think why anyone needs it. If you're making a drawing
> program, you should use core X to do the GUI graphics, and OpenGL to
> do the actual drawing-widget contents.

hell no. i wouldnt bank my app on opengl for that. i'd make an abstraction and
then slide in whatever engine u want under that to do the actual grunt - from
software rendering in client-side buffers and then xshmputimage the bastards
across, to xrender, to opengl, etc.

opengl is quite "iffy" for 2d - especcially if u want to use textures on pilys
for replacements of pixmaps - different ogl implementations will round texture
co-ords diddfernly so u may not end up with precisely the same output on every
driver - and sometimes the output is less than desireable.

-- 
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler)    raster at rasterman.com
裸好多
Tokyo, Japan (東京 日本)