On Saturday 11 March 2006 02:08, Eddy Petrişor wrote: > > AFAICT, that would be the range from 10A0-10FF [1]. > > For now, I'd propose to include the full range. > > Already done (see r35369) Although it is not enterly clear from the comments in the script, I assume this script is intended to do the stripping at d-i build time. I do appreciate your work Eddy, but, as I've said before, I'm not in favor of this and thus it's not going to happen unless someone can give very good arguments why we should. Here are my reasons for not doing it. - Stripping at d-i build time has the advantage that we keep it in our own hand and that we could, in theory, strip based on actual glyphs used. - Stripping at build time also means that d-i would get yet another build dependency (fontconfig) and that the stripping will need to be done for each and every build for every arch supporting g-i for something that is in the end relatively static. These disadvantages IMO outweigh the advantages mentioned in first point. - There is a lot less need to do extreme stripping of fonts for g-i than there is for d-i as the required memory for g-i is a lot bigger anyway and the relative saving by extreme stripping is not significant. So, we should definitely do stripping of fonts to avoid overlap, but we should also do it in the udebs themselves because the stripping is relatively static and not at d-i build time. Cheers, FJP
Attachment:
pgpspu3WMUV4S.pgp
Description: PGP signature