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

Re: #338638 (tetex-base: dies when /usr/local/ read-only)



Hilmar Preusse <hille42@web.de> wrote:

> Hi,
>
> That bug is actually closed but not yet fixed in stable. The last
> upload to stable was rejected by Martin IIRC.
> Now, as he's not willing to maintain stable any more: Should we try
> to get it into stable a second time in the hope the new maintainers
> won't reject it?

Yes, but IIRC there was still a flaw in the package.  This wasn't the
reason for the rejection, because I only discovered it myself when
thinking about bringing this topic up on -release, after my private
conversation with Joey lead to the rejection.

Okay, I digged it up.  Here's the last mail I wrote to him, in german,
with some english comments:

,----
| vorweg: Ich stimme dir zu, dass es kein update von tetex-base in stable
| geben sollte; aber aus anderen Gründen.
| 
| Ich halte es mittlerweile nicht mehr für sinnvoll, weil mir klar
| geworden ist, dass es das Problem nur verschieben, aber nicht lösen
| würde:  Es würde dann der upgrade von 3.1 zu 3.1r1 scheitern, anstatt
| der von sarge zu etch.  Das wäre zwar etwas leichter zu handhaben, aber
| das ist es nicht wert.

If the upload had been accepted, the way it was made the upgrade between
sarge point releases would have failed instead of the dist-upgrade to
etch.  So to do it properly, the fixed package targetted at sarge must
have a proper code for the failed-upgrade case in its prerm (dpkg
calling "new-prerm failed-upgrade <old-version>"), which I left out
erroneously (it doesn't help to put this into the etch packages, because
tetex-base will be removed before the new tetex-bin is unpacked, so
there's no "new-prerm").

Now come the reasons why I think it should be accepted into stable:

If a package in stable...

| Generell glaube ich aber, dass wir schon noch einen Dissens haben:
| 
| Wenn ein Paket in stable 
| 
| - einen Bug hat, der nur beim (dist-)upgrade zum
|   neuen Stable getriggert wird, 

has a bug which is triggered upon (dist-)upgrade to the new stable [I
wrote "only", but that's wrong, removal too]

| - der sich nicht durch Änderungen an Paketen für das neue stable lösen
|   lässt, 

which cannot be fixed by changing packages in the new stable

| - wohl aber durch ein update von stable, und

but can be fixed by a stable update

| - der zum Abbruch des (dist-)upgrade führt,

and which leads to a failure of the (dist-)upgrade

| dann finde ich, dass das ein Kandidat für ein update in stable ist.

then I think it's a candidate for stable.

| Natürlich mit entsprechender Abwägung (wie groß ist der Eingriff in
| stable?  Sind die Folgen beim dist-upgrade gravierend, d.h geschieht der
| Abbruch an gefährlicher Stelle oder lässt er sich trivial lösen?
| Wieviele der user des Pakets sind betroffen?).
| 
| Bist du da anderer Meinung?  Ich finde, dies sind wir unseren Usern
| schuldig. 
`----

Regards, Frank
-- 
Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX)



Reply to: