Re: Bitmap objects

[Pete French]

> Actually that is not missing.  The data sources should
> be considered as a continuous bitstream with highorder bits first.
> However, every row of te source image starts at a character boundary.

So the image operator doesnt natively handle images with padding at the end
of a row. Makes sense actually (and, to be honest, I dont think I have ever
come acros such an image "in the wild" as it were).

> Second, I can't remember seeing a colourspace in the xgps backend
> at all.

>From looking at the code it appears to be in color.space and is set
by DPSsetgray, DPSsetrgb and DPSsetcmyk.

I decided that the best way to handle this was to actually set the
colourspace inside NSDrawBitmap and then use whatever the current
colourspace is within the bitmap compositing function.

> However, the image dictionary actually has a required parameter
> `Decode' which is an array and contains 2 values per (colour) component.

I was intending to ignore that to be honest - the only use it would have in
normal operation would be for one-is-black greyscale images, which is probably
better handled by a flag added to the dictionary. Faster too. One thing 
dont want to do is add loads of general case code that ends up slowing down
the rendering of bog-standard 8 bit packed images that are used for icons
and the like.

> There is the parameter `Interpolate' that defaults to false
> which might be used to switch between fast / faithfull.

Its only really an issue with minifaction (for magnification it is
acceptable OpenStepy behaviour to simply magnify the pixels) I'm going to
try it several ways and compare speeds - readble minified images are nice to
have for minimised windows after all.

-bat.