Disttags are nice, save the disttags

[Axel Thimm]

On Tue, Jun 05, 2007 at 11:35:24AM +0200, Patrice Dumas wrote:
> On Tue, Jun 05, 2007 at 09:56:20AM +0200, Axel Thimm wrote:
> > On Tue, Jun 05, 2007 at 09:32:04AM +0200, Patrice Dumas wrote:
> > > On Tue, Jun 05, 2007 at 02:51:36AM +0200, Axel Thimm wrote:
> > > > 
> > > > python on different Fedoras have different ABIs and different module
> > > > installation paths, so even if a python noarch module you have to
> > > > rebuild python modules from FC6 (2.4) to F7 (2.5).
> > > 
> > > Sure, but in my recalling this happens less often than changes in the
> > > C/C++ build chain, so it may be more worth not having disttag to limit
> > > unneeded updates.
> > 
> > It's about yearly or every other Fedora release. Or twice during the
> > RHEL release cycle. We do want to support upgrading from FC<N> to
> > FC<N+2> starting with N=5, so it is definitely within our range.
> 
> It has advantages, but in my opinion this is clearly a case where the
> balance is not completly self evident and should be left to the packager
> (it should always be left to the packager, but the balance may be on one
> side or the other more clearly in other cases).

Well, that's the case, disttags were never enforced, they made it into
the majority of packages (89% with a 10% increase on each release for
th epast 3 releases) because they are useful.

If the maintainer of a package wants to use an integer cache in his
basement, he's free to do so. He'll just get virtually kicked in the
ass when his manual and error-prone procedure fails, until then
everybody and his rabbit is happy.
-- 
Axel.Thimm at ATrpms.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/fedora-maintainers/attachments/20070605/19ff113a/attachment.sig>