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

Re: APT overrules self-compiled packages

Colin Watson wrote:
> On Mon, Aug 12, 2002 at 01:04:14PM -0500, Jamin W. Collins wrote:
> > Yea, but I think it would make more sense for the new package to be
> > created as more of a fork rather than an incremented version.  I ran into
> > this myself a while back and now that I've read through more of the
> > maintainer and policy documentation it seems that it would make more sense
> > to create a new package name that "provides" the old package name.
> Provides currently can't satisfy versioned dependencies, so this will
> often be inadequate.

Neither can incremented version numbers in all cases. Consider:

Depends: foo (= 1.0-1)

This is a prime reason why this behavior of apt stinks. See exactly this
trouble being run into by the apt-build program in bug #155170. I wish this
behavior could be turned off by easier means than setting up your own
apt repository.

By the way Julien, could you explain what you did to close bug #155170? I need
to add some similar fix to apt-src probably.

> Personally I think   it's a good idea, since it makes bug reports less
> confusing in the event of self-compiled packages, and helps me
> remember which packages on my system I built locally and which are
> official.

Of course neither of these benefits accrue if you go ahead and set up a local
apt repository for locally built packages, something that apt's current
behavior inclines me to do. Hmm, now that I re-read apt-build's changelog, 
I think it goes so far as to do that for you, all to avoid this misfeature of 
apt. Crazy!

see shy jo

Reply to: