Re: libgeda & soname changes (was Re: ongoing curl transition pain)
On Fri, Jun 22, 2007 at 12:50:55AM +1000, Hamish Moffatt wrote:
> On Thu, Jun 21, 2007 at 12:42:29AM +0100, Steve Langasek wrote:
> > On Wed, Jun 20, 2007 at 11:40:35PM +0200, Domenico Andreoli wrote:
> > > uploaded, sitting in the NEW queue.
> > Now in incoming. :)
> <Grumble> I don't suppose there is any chance of the rest of NEW being
> looked at?
At some point, yes. But don't expect the release team to ask for expedited
NEW processing on your behalf for an soname change that will /add/ to their
current workload in britney, when we already have things like the curl
transition blocking most packages from reaching testing today.
> - Therefore I don't see any value in having the correct soname in
> the library's binary package name. This just creates delay and
> work for ftp-master for no return.
Using versioned package names is still beneficial even within small package
groups because it allows libs to be coinstallable on upgrade and simplifies
the upgrade calculation for frontends such as apt/aptitude vs. the versioned
conflicts approach you used previously. Even if you don't perceive a
benefit here for your own packages, there is a cumulative benefit of
reducing the number of dependencies of this sort in the release overall.
> We discussed this in #400307 before etch was released. At the time I
> fixed the dependencies but left the soname wrong. This violated policy
> but was not actually harmful.
> Now with the latest upload I fixed the soname, and have immediately hit
> delayed NEW processing again. This was the reason why I stopped updating
> the library name a while back.
> What do you suggest?
My suggestion remains unchanged. I'm sorry that NEW processing isn't
consistently speedy, but again, at the moment some lag time there is helpful
to the release team anyway, not harmful...
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.