Since you're using the new immVertex2s function, should be:

unsigned pos = add_attrib(format, "pos", GL_SHORT, 2, CONVERT_INT_TO_FLOAT);

or

unsigned pos = add_attrib(format, "pos", COMP_I16, 2, CONVERT_INT_TO_FLOAT);

If you build with "make debug" or "make lite debug" or (on Windows) "make 2015 lite debug" it will catch things like this at runtime.

same here

GL_FLOAT (or COMP_F32), 3, KEEP_FLOAT

GPU_SHADER_3D_UNIFORM_COLOR

Debug build will catch things like this. Would it be better if I enabled more error checking for non-debug builds too?

Call immThemeColorShadeAlpha after immBindBuiltinProgram(GPU_SHADER_3D_UNIFORM_COLOR).

Same as above, use 3D versions of these.

Use imm_cpack here.

My bad. Still getting used to the APIs.

I can't find a definition for immThemeColorShadeAlpha().

They're new to everyone, it's expected!

Debug and release builds will now catch errors at runtime.

Oops! immUniformThemeColorShadeAlpha