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

Re: additional virtual packages for kde



Christian Schwarz <schwarz@monet.m.isar.de> writes:

> Now to the critical point: I don't think it's our job to change our
> packages in such a way that some third party (hear: KDE) can replace them.
> In fact, I think it's the third party's job to fix _their_ packages so
> they integrate into the Debian system well and don't allow mixing with our
> corresponding packages. 
> 
> Here is the reason for this: By changing our packages so that they are
> compatible with packages from someone else we would give away some piece
> of our freedom and independence. 

I agree.  This is very slippery slope.  While I haven't yet decided
what I think about the KDE case, we have to be very careful when it
comes to letting outside projects (indirectly) modify our packages.

(This next bit is somwhat off the KDE topic, but I think it's an issue
worth considering.)

Consider this (much more extreme) scenario: Microsoft compiles their
Interest Exploder for Linux and releases it as a Debian package[1].
In their postinst they do something which trashes the mime-types
database, claiming it's fine since IE has no problems with it.  Do we
then go so far as to create a debian package, "ie-repairs", which
fixes the problems, or do we just say "their package is broken", don't
use it?

This situation makes me think we might eventually want a database
which can be used to list "problem" packages.  dpkg would refuse to
install any packaage whose name glob-matched a line in the database
unless the user uses something like --force-problem-packages.  We
could then just list IE there:

  # /etc/dpkg/problem-packages
  # Package-name         Version 
  
  internet-explorer*        *

(I also thing the "Origin:" idea is probably a good one, but it's too
general for many problems.)

[1] Of course, we know they'd release it with "Provides:
microsoft-browser" and "Conflicts/Replaces: <every-other-browser>".
But that's a different sort of problem :>

-- 
Rob Browning <rlb@cs.utexas.edu>
PGP fingerprint = E8 0E 0D 04 F5 21 A0 94  53 2B 97 F5 D6 4E 39 30


Reply to: