Bug#244601: tetex-bin: package fails to configure: Error: `mpost -ini -jobname=metafun -progname=mpost metafun.mp' failed
Frank Küster <frank@debian.org> wrote:
> Branden Robinson <branden@debian.org> wrote:
>
>> On Fri, Apr 30, 2004 at 02:03:54PM +0200, Frank Küster wrote:
>>> Branden Robinson <branden@debian.org> wrote:
>>>
>>> >
>>> > Okay, I've attached a typescript and the output of ps ax.
>>>
>>> So it's either fmtutil.cnf that's missing, or the symlink from
>>> /usr/share/texmf/web2c/ to /var/lib/texmf/web2c.
>>
>> It appears to be the latter.
>>
>> $ ls -dl /var/lib/texmf/web2c /usr/share/texmf/web2c/
>> drwxr-xr-x 2 root root 4096 Apr 26 12:54 /usr/share/texmf/web2c/
>> drwxr-xr-x 2 root root 4096 Apr 26 12:54 /var/lib/texmf/web2c
>>
>> Hmm, that's bad. Both are plain directories, and neither is a symlink.
>>
>> You might need directory-to-symlink migration logic like I have in XFree86.
>
> Hm, yes, for upgrades from potato probably. I'm not sure, we'll have to
> check the potato package and check this. But it seems we need to check
> this if we missed it upon the potato-to-woody upgrade.
No, that was a symlink even in potato. I don't know anything about slink
and earlier (TEXMF must have been in /usr/lib/texmf long ago), but at
least this is not a transition problem from potato or woody to sarge or
unstable.
What strikes me is that you encountered that problem exactly
now. tetex-bin cannot work properly without proper contents in
/usr/share/texmf/web2c/, it will at least fail each time postinst is
called.
And you said that this machine has been "steadily upgraded since potato"
- this means that it was first installed with potato?
How long was it, approximately, since the last upgrade before this one
that caused the error? Could it be that the link got somehow lost, and
you created the directory /usr/share/texmf/web2c manually, because this
solved some problems you encountered?
> Can you point me to where exactly you use this directory-to-symlink
> migration logic - I guess in one of the preinst scripts?
It seems we don't need this.
Would you agree that we downgrade the severity of this bug? Your system
is inconsistent wrt tetex in a way that will stop it working completely;
and I can see no indication that any recent package of ours has caused
this. It seems it needs really weird, improbably co-occurrences to
trigger this, if not manual intervention. On the other hand, you
encountered the bug, and I trust you to be someone who knows what he's
doing, especially wrt "manual intervention". So I won't want to change
anything unless you agree.
Regards, Frank
--
Frank Küster, Biozentrum der Univ. Basel
Abt. Biophysikalische Chemie
Reply to: