Re: Removing assumption of unsigned long pixel values for colours

[Alex Gramiak <agrambot@gmail.com>]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Alex Gramiak <agrambot@gmail.com>
>> Cc: emacs-devel@gnu.org
>> Date: Mon, 06 May 2019 10:29:38 -0600
>> 
>> > COLORREF is just the RGB representation of a color, see
>> >
>> >   https://docs.microsoft.com/en-us/windows/desktop/gdi/colorref
>> 
>> Right, but my point is that AFAIU the unsigned long value is used in a
>> different manor (embedded rather than used to lookup a value) on the w32
>> side, so it could be viewed that the representation of colors is already
>> platform-dependent.
>
> OK, but that just says that each platform already performs the mapping
> you wanted to avoid.  What makes your back-end special that it needs
> to avoid this?

My initial reasoning was the time argument, which turned out to be
wrong.

Now I want to avoid the lookup table approach and try to just embed the
GdkRGBA into a pixel value, but unsigned long is not necessarily wide
enough to support this.

Should I take the lack of response to the other point as an OK for using
the emacs_pixel typedef approach? An unsigned long long would be enough
to fit the 16bpc ARGB value in.