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

Re: fonts in user home



Frank Küster <frank@debian.org> wrote:

> Given that it is quite unlikely that root will have his own
> $TEXMFCONFIG, I prefer the id check.  For systems where a

OK.

> non-priviledged user has just enough rights to run update-updmap (and
> updmap-sys), we can add an optional "--output-dir=..." parameter.

Mmmm, do you really think this kind of thing is going to happen?
Currently, update-updmap forces root:root and 644 mode for updmap.cfg,
but in the setup you are describing, that shouldn't happen in
system-wide mode (which is the only mode currently existing in
update-updmap). We would have to somehow guess the correct
permissions---I don't want to do that---or leave the 600 mode that
mktemp chose (which would be bad for users). And to get the owner group
right, I would have to create the temporary file in the destination
directory instead of /tmp... and to trap signals to make sure it is
always deleted. But which signals? Not all of them are standard. tetex
scripts trap some of them, but I don't know why they trap these signals
and not others, and that is why I didn't put such a trap in
update-updmap...

Actually, I was thinking the other way round: crazy people really
wanting to work as root, and therefore wanting the user-specific
behavior or update-updmap (looking under $TEXMFCONFIG/updmap.d/ for
additional .cfg files and writing, at least by default, to
$TEXMFVAR/web2c/updmap.cfg[1]). For them, I could add a --user-specific
option to override the id check, but maybe this is also too braindead to
consider.

[1] Ah, this is not that simple. Theoretically, TEXMFCONFIG and TEXMFVAR
    can expand to several directory trees, right? Therefore you cannot
    use (for instance) "kpsewhich --expand-var '$TEXMFVAR'" to get a
    directory name to write to... We could check whether each of them is
    a directory before using them, and tell the user to speficy the
    directories with two new options otherwise...

-- 
Florent



Reply to: