🚀 go-pugleaf

I have application that I've been working on for a while that uses NO
palettes. It can read and write 16 color .BMPs fine. Now I'm adding palattes
so that the application can handle 256 color .BMPs. I have the display working
fine and the "paint" routines etc. I can read the .BMP in, add it's colors
to my palette and "directly" send it to the screen with SetDIBitsToDevice.
Sending to the screen directly was only a sanity check. I really want to
copy it to a particular location within the backing-store bitmap. If I bitblt
it into the backing store I only get 16 colors.

What is the proper procedure for copying a small DIB into a larger one
and keeping the palettes straight. I've tried using 

BitBlt (only 16 colors, not surprised)
SetDIBIts (some thing show up, compressed into a couple scan lines)
SetDIBitsToDevice (works directly to screen, nothing shows copying to DIB) 
StretchDIB (same as above)

To add the colors to the palette I loop through .bmicolors[0..n] and add
them if not already in the palette.