[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#478704: texlive-xetex: 478704: fontconfig instead



On Sat, 2009-12-05 at 15:27 +0100, Norbert Preining wrote:
> On Sa, 05 Dez 2009, Paul Wise wrote:
> > Yeah, they aren't particularly well standardised. I don't think it
> > matters much which you use.
> 
> Hmm, is there no FHS or policy on that?

Not sure about that.

> > You add the symlinks now and then not worry about the fontconfig stuff
> > at all. Xorg will probably get your fonts when #539338 is fixed, the
> > patch needs testing/checking though, not sure if it follows symlinks.
> 
> The discussion on the bug report shows that it goes down 100 levels
> of dirs including symlinks.

Cool. That should be enough levels too.

> > Install all your fonts to /usr/share/fonts instead of the current paths
> > and then not worry about the fontconfig stuff at all. Xorg will get your
> > fonts when #539338 is fixed.
> 
> No, I don't want to create links for each and every font.

Why would you need to do that?

> > File a bug on fontconfig to get your font paths added to its
> > configuration and triggers. Also file a bug on Xorg to get your font
> > paths added to the default FontPath. The latter is needed to get your
> > fonts into Xorg during the period where it doesn't support fontconfig,
> > which will be quite a while.
> 
> Hmm, again I am a bit puzzled.
> 
> As far as I see there are two things involved:
> 
> 1. fontconfig accessing the fonts
> 2. X accessing the fonts
> 
> For 1. I can simply wait until fontconfig is trigger-enabled, at which
> point the fonts will be available to fontconfig immediately.

fontconfig is already trigger-enabled, but it doesn't know about the
non-standard path TeXLive uses for fonts.

> For 2. we need mkfontscale be recursively and trigger enabled, as
> discussed in bug #539338, but that seems to take ages, right?
> But when I put the fonts below truetype and type1 in /u/s/fonts
> they are already in the default font path, so the only thing
> that would be needed is fixing mkfontscale, and there is already 
> a bug.

Right.

> So for what I see there are two options. In both cases I drop
> links below /u/s/f/truetype and /u/s/f/type1.
> 
> Then
> A. I call fc-cache in every font shipping package
> B. I do NOT call fc-cache ...

You should do B....

> Doing A. would do 1. from above, while doing B. would mean that we
> have to wait for font config triggers to be implemented to get
> fontconfig accessing the fonts

...because fontconfig already has triggers for /u/s/f/

> Regarding 2. X accessing the fonts there are again some options:
> - short term solution:
>   register all the fonts manually with X
> 
> - mid term solution:
>   fix #539338 and the fonts will be available
> 
> - long term solution:
>   Xorg gets fontconfig support built in

Right.

> The only thing I can take is the short term solution, but I will not 
> do that, since that are too many fonts and I don't want to do that
> for all the directories.

Fair enough.

> So my proposal is: link as described above and do nothing else for now

Sounds fine to me I guess.

I guess moving all the fonts to the standard /u/s/f/{t,o}/ directory is
out of the question?

> Effect: fonts will not be available to X, and to fontconfig only after
> someone else has called fc-cache.

The fonts missing from X isn't a regression though, so that is fine.

fontconfig will call fc-cache from the trigger in its postinst, so the
fonts should be available to fontconfig-supporting apps immediately.

> After that is in unstable and we have fixed the billions of RC bugs 
> coming in after the first upload to unstable, I will try to add
> the fc-cache calls to the font shipping packages' postinst and postrm
> scripts, which will make the fonts available automatically for fontconfig.

That isn't needed because of the existing fontconfig triggers.

> On the X front I wait for the above to be fixed, or any other solution,
> and do nothing.

Correct.

-- 
bye,
pabs

http://wiki.debian.org/PaulWise

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: