On 06.04.05 Steve Langasek (vorlon@debian.org) wrote:
> On Wed, Apr 06, 2005 at 01:33:19PM -0700, Steve Langasek wrote:
Hi,
> > Reassigning to jadetex accordingly for further investigation;
> > hopefully this is a known bug in previous versions of jadetex
> > that have since been fixed.
>
> ... nope, guessed wrong here. Reassigning back to tetex-bin.
>
> This bug has been fixed in tetex-bin 2.0.2-28:
>
> tetex-bin (2.0.2-28) unstable; urgency=low
>
> * The maintainer scripts now will again fail if the format or map file
> generation fails. This has been turned into a debconf note in version
> 2.0.2-19 as an emergency action, and must be dropped now the packages
> proved to be stable.
>
> This might reveal some bugs that had been hidden so far, and might
> help debug #302827. [frank]
> * In postinst, fix language.dat if it points to an old french
> hyphenation file that we no longer ship, and which only loaded the new
> one even in woody, thanks to Sven Koch <haegar@sdinet.de> (closes:
> #302286) [frank]
>
> -- Frank Küster <frank@debian.org> Mon, 4 Apr 2005 20:41:26 +0200
>
> Frank, this install failure was *caused by the use of debconf*.
> The debconf-devel(7) explanation for INPUT (db_input) says:
>
> Debconf decides if the question is actually
> displayed, based on its priority, and whether the
> user has seen it before, and which frontend is being
> used. If the question will not be displayed, debconf
> replies with code 30.
>
> You were not checking the return code of db_input; therefore, the
> maintainer script was exiting with code 30 on the buildds.
>
> The actual format creation failures all seem to have been caused by
> stale config files on the system from previous versions of jadetex;
> but the tetex-bin install failure was caused by this misuse of
> debconf.
>
OK, great.
> If you are going to readd the debconf note at a later date, please
> make sure you fix this usage problem. (If what you really want is
> to force the display of the debconf note, you should be using
> db_fset seen.) If you are not going to readd the debconf note in
> the future, then this bug can be closed.
>
That thing was introduced in -19:
tetex-bin (2.0.2-19) unstable; urgency=high
* If the generation of formats (and map files) fails in postinst,
display a critical debconf note, instead of exiting with an
error. This will lower the severity of some bugs, where the
failure is due to a user not accepting conffile changes.
Furthermore, the warning displayed with the last critical
conffile change is now more "convincing". (Closes: #266497).
[frank]
-- Frank Küster <frank@debian.org> Thu, 19 Aug 2004 17:06:08 +0200
and we were sure, that it would be only a temporary solution. It has
been reverted now. So I guess that bug can be closed now. Actually
tetex-bin and jadetex install fine:
http://buildd.debian.org/fetch.php?&pkg=pdns&ver=2.9.17-9&arch=powerpc&stamp=1112740284&file=log&as=raw
Regards,
Hilmar
--
sigmentation fault
Attachment:
pgpsDh6b5eVxa.pgp
Description: PGP signature