Re: postinst and abort-remove/deconfigure

[Frank Küster <frank@debian.org>]

Norbert Preining <preining@logic.at> wrote:

> On Die, 11 Okt 2005, Frank Küster wrote:
>> >   - you appear to think nothing should be done on disappear? I think the
>> >     package gets removed and therefore, the map files should be updated
>> >     accordingly;
>> 
>> I think that also the teTeX packages do not take into account that
>> postinst can be called with abort-remove or abort-deconfigure, and
>> similarly for other maintianer scripts.  Now with texlive we should.
>
> Two questions:
> 1) Do we have to take "disappear" into account for TeX packages? We 
>    conflict with each other, so we will for sure never replace each
>    other. AFAIU this is the precondition for a disappear call?

I'm not sure that we won't still replace each other.  But in fact we
should care for the "in favor" arguments.  Or in other words, before the
texlive packages are uploaded to unstable, we should

- first test installing them over an installed tetex system, and check
  which changes are needed in both packages

- once that works, test the other way round.

> 2) (ignorant Norbert) What would be on the todo list on case of a
>    abort-remove and abort-deconfigure? No idea for me.

I don't have any TODO list currently, either.  But I just noticed that I
never cared about these arguments (simply because they weren't in the
maintainer scripts I inherited), and I figure they might occur now that
we have texlive.  I didn't really think about it thoroughly; the mail
was more a reminder.

Regards, Frank
-- 
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer