Re: [AD] About kryten2's bug

["Eric Botcazou" <ebotcazou@xxxxxxxxxx>]

> I propose to let best_fit_color() and RGB mapping tables treat index 0 as
> any other index, so that makecol8() does the same.

Finally, I think that's not a good solution because bestfit_color() and RGB
mapping tables are used to build various other color maps and I don't want
to break everything.
So index 0 will still be handled apart from other indexes. I simply propose
to even the behaviour of bestfit_color() and that of RGB mapping tables.

> Similarly, I propose to change the behaviour of get_optimized_palette() so
> as to make it build 'adapted' palettes only if the KEEP_TRANS flag is set.

Same thing, finally not very good because we would end up with two types of
palette. So default palettes will still have index 0 set to pink color.

I've attached a third patch (color.diff).
If no one disagrees, I'm going to commit grabber2.diff, quantize.diff and
color.diff and close krysten2's bug on SourceForge's tracker.

--
Eric Botcazou
ebotcazou@xxxxxxxxxx

Attachment: color.zip
Description: Zip compressed data