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

Re: Packages with "fc6" name in Fedora 7



On Wed, Apr 11, 2007 at 09:04:59AM -0400, Jesse Keating wrote:
> On Wednesday 11 April 2007 08:31:15 Axel Thimm wrote:
> > The fc6 tag is really cosmetics in comparison to what we may run into
> > w/o a proper rebuild. In fact we should really keep them (the tags),
> > so when a package explodes the user/bug reporter/bug assignee will be
> > able to identify the distribution the package was built on and perhaps
> > derive that that's the real issue.
> 
> With Koji, it is pretty easy to find out EXACTLY what packages were in the 
> buildroot to build any given package.  The dist tag is not necessary for 
> guessing.

We did have buildlogs with plague, too. I don't know how long they
were kept, though, and certainly no user/reporter ever cared to look
into them.

> All your above scenarios are valid, and can be mitigated by on the side 
> continuous rebuilds of packages to identify when changes might happen, which 
> would allow the maintainers and release team to make decisions as to which 
> packages should be rebuilt for a new build chain.

You only test whether it builds, not whether it runs. To continue the
bridge-utils example: If it is busted and only shows that it is once
you try to setup STP on the bridge, who will teh continuous rebuild
show you that? You can only do that by either investing in-house QA
for every package update you are going to ship (and we know that these
are going to be a loadful per week), or do so in advance during the
development/testing cycle.

It's just a matter of when it will happen, not if and how much
resources it will consume. And I think having it happen during F7testX
is better than during F7 package maintenance, or even worse, if it
slips the package updating QA and makes it into the proper released
updates. Because I doubt that a one-line fix in the ever abused
bridge-utils example will make the packager that fixed it to test all
aspects of Linux bridging again.

> It does not mean we should just blanket rebuild everything just
> because we can and there could POTENTIALLY be issues in SOME
> packages.

If the issues are only potentially, why are you afraid of rebuilding?
The arguments were a) too much download volume and b) stable
packages. 

The first argument is void, since the big players consuming 99% of the
bandwidth have been rebuilt (for the record, w/o weighing by size FC6
had 80% rebuilt, and the big players were among them).

The second argument on stability is also void, since either the
package is stable and survives a rebuild, or it is as fragile to not
do so. And in that case, we'd like to know before the release, that
the package is in a fubar state of affairs.

You don't save anything, you just push the problem from the development
cycle into the maintenance cycle.
-- 
Axel.Thimm at ATrpms.net

Attachment: pgpNpfONquT1O.pgp
Description: PGP signature


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