For your consideration: Secondary Architectures in Fedora

Daniel Veillard veillard at redhat.com
Thu Jun 14 10:06:49 UTC 2007


On Wed, Jun 13, 2007 at 03:20:33PM +0100, David Woodhouse wrote:
> On Wed, 2007-06-13 at 10:11 -0400, Daniel Veillard wrote:
> >   yeah yeah, and s390x pointed out a bug in libxml2 about some obscure
> > aliasing problem. But in general going though that wasn't very fun.
> 
> Yeah, finding bugs isn't fun. Let's never increase our test coverage.

  haha, nice rethoric. The problem is *when* do you test ?
Compiling after release time on the weird platforms is nearly the
worst time possible to run those checks. That means the code is
already upstream, you're usually under tight timing constraints
for pushing, and blocking your build because say Ekiga doesn't built
on s390 puts the packager in the most awkward position possible.
He gets to try to fix a bug raised on an architecture he has no knowledge
of, without help from the upstream community, blocking everything else
and then forcing him alone to come with a "quick fix". A sure recipe
for disasters ...
  Suggesting I'm against testing software portability is a good joke,
thank you, I'm sure the whole libxml2 community will have a laugh if
you forward it in the proper place :-)

  Now to get back to serious stuff, yes we cleaned up most of the packages
which constituted Core, but we are gonna regress now because compilation
on the 'odd' platforms will now occur even more asynchronously when the
Red Hat employees will start to recompile for RHEL release, possibly one
year after the software has been pushed out. The later you catch stuff
the more expensive it is, I doubt anybody will disagree, right ? We don't
want to have those weird platform in the way of Fedora apparently (I 
don't know the exact reasons). Can we have them in parallel ? I'm thinking
of asynchronous builds of what constitutes Fedora done on those box and
error reports being raised to the packager or as bugzilla entries if they
fail. That would be one start.
  But blocking the packager synchronously at build time is really not
proper IMHO.

Daniel

-- 
Red Hat Virtualization group http://redhat.com/virtualization/
Daniel Veillard      | virtualization library  http://libvirt.org/
veillard at redhat.com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine  http://rpmfind.net/




More information about the fedora-devel-list mailing list