Still not understanding pixmaps vs RGBs

["Lewin A.R.W. Edwards" <larwe larwe com>]

Sorry, I'm probably missing something major, but I don't see what 
pixmaps gain me and I'd be really appreciative of some more explanation. 
Currently my application works like this:

* Render content into an RGB buffer (Still image decoding and some 
special postprocessing). This buffer is the size of the source image, 
not the screen size.
* Decimate or expand content, using my own hand-rolled code, into a 
second RGB buffer that is screen-sized.
* Draw the RGB buffer into a darea that covers the whole screen (this 
step is slow).

One of the reasons I wanted to move to pixmaps is that someone here 
recently told me the pixmap functions support resizing. However, they 
don't. gdk_draw_pixmap is a 1:1 copy (possibly with clipping). Are there 
functions, maybe in XLib, that will do pixmap resizing (scaling) for me? 
My decimation/expansion code is horribly inefficient, I think.

One of the other reasons I wanted to move to pixmaps is that they are 
the same memory layout as the screen, which is 16bpp - so I'd have 30% 
less data to move around compared to the same sized RGB buffer. So I 
wanted access to the internal structure of the pixmap, on the assumption 
that it was internally a header (dimensions, depth, phase of the moon 
information) followed by a flat display-format buffer.

Is a pixmap something like a PICT on the Macintosh, i.e. more like a 
metafile format than just a frame buffer? Or is it like a 
device-dependent bitmap on Windows?

--
-- Lewin A.R.W. Edwards (Embedded Engineer)
Work     http://www.digi-frame.com/
Personal http://www.larwe.com/
Tel (914) 937-4090 9am-6pm ET M-F