> It really wouldn’t surprise me too much if this was all related to
> masks, I never managed to get a satisfactory test going.
> 
> OTOH, if XPMs don’t even use these functions then that would certainly
> cause scaling to fail. I’ll have to have a look at the XPM code to
> find out what they’re doing instead.

OK, so when there’s a mask there’s a different function called in
xterm.c to draw the image. I’m not sure what the deal with that is but
I had to modify it to use x_composite_image. We also need pretty much
every other patch that we’ve posted to this thread, so I’ve attached
something that works for me.

I want to go back and add a couple of comments to it, so it’s not
final, but can you please test it and see if you can break it.

XPMs are still not scaling. I suspect PBMs are in the same boat. It
looks like the code for them creates the pixmaps themselves instead of
using image_create_x_image_and_pixmap, so they bypass the whole
XRender Picture creation. It should be relatively easy to just add in
the creation of a Picture.
-- 
Alan Third