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

Re: Braindump: Can we get rid of the font-cache-group question?



On Wed, May 24, 2006 at 13:11 +0200, Frank Küster wrote:
> Norbert Preining <preining@logic.at> wrote:
> > On Die, 23 Mai 2006, Ralf Stubner wrote:
>
> I think we should try to find a solution where font cache data do not
> usually disappear regularly.  Either a per-user setup in $HOME, or a
> system-wide setup in /var/cache/fonts.  /tmp should only be a last
> resort. 

ACK

> >> But old cached font
> >> data would still be lost on normal user machines.
> 
> But only on machines where TEXMFVAR is not writable, which doesn't sound
> very probable.  Therefore, I think this is not an issue.

I meant cached font data that is *now* in VARTEXFONTS. That will be
'lost' if VARTEXFONTS is changed. As you mention correctly later, this
could be considered a feature, since it increases the possibility that
users know about this change. This would be good not only on multi-user
machines, where the admin might want to set up a shared font cache, but
also in the general case, as a user might want to change exclusion rules
for backups etc.
 
> Couldn't we do it *without*any* questions?
> 
> - /var/cache/fonts exists, with standard permissions (0755)
> - mktex.cnf is a conffile, with 
> 
>   VARTEXFONTS=/tmp/texfonts

Shouldn't VARTEXFONTS be still defined in texmf.cnf?

>   MT_FEATURES=texmfvar
> 
> - NEWS.Debian and README.Debian instruct how to set up a system-wide
>   VARTEXFONTS directory in /var/cache/fonts or elsewhere
> 
> * new install:
> 
>   Fonts are cached in TEXMFVAR, except on strange setups where it is not
>   writable.  For single-user machines, this has functionality as good as
>   caching in /var/cache/fonts.  For multi-user machines, people can read
>   in README.Debian how to centralize the font cache
> 
> * upgrade:
> 
>   Fonts in /var/cache/fonts are lost (this is even better than
>   README.Debian, people will simply notice it...) unless the admin
>   changes settings.

Sounds good, at least if we can get the problem with nonexisiting
TEXMFVAR covered.

cheerio
ralf



Reply to: