Packaging guidelines: IPv6

[Ralf Corsepius]

On Wed, 2006-07-05 at 09:28 -0500, Jima wrote:
> On Wed, 5 Jul 2006, Ralf Corsepius wrote:
> > It's even worse: All FE currently has is an "initial this looks sane
> > enough" review. Once a package is in FE, there actually is no QA nor
> > audit on packages at all. Nobody but the package owner is allowed to
> > change packages. If he doesn't want to listen, nothing will happen,
> > maintainers have all kind of freedom to commit all kind of stupidities
> > they want.
> 
>   Huh?  If a packager is committing "all kinds of stupidities" (catchy 
> phrase, I like it) to their packages, someone's bound to notice in the 
> emails to fedora-extras-commits (unless the packager is *really* good at 
> hitting ^C at the right times!).
How many people are reading fedora-commits? I guess very few ...

I read it "batch style" every now and then and occasionally comment on
some (from my POV) eye-striking "oddities" ... In many cases, packagers
prefer to ignore such comments and to remain silent.

>   If not, someone might notice bizarre 
> behavior in the package and eyeball the CVS. 
Well, in many cases the "bizarre behavior" is noticed by the public,
when such a bug hits the "common man".

Classical examples having hit fairly frequently would be
* "SONAME changes"
* broken NEVRs.
* Shipping immature/premature beta/alpha stable SW.

With a little carefulness at least the former 2 cases would be
avoidable, and with a little team spirit, they in many cases would often
be easy to fix. 

So far this apparently isn't the case. Instead we are watching mails
concerning "broken deps", and every now and then are facing the
consequences of SONAME changes (rebuild requests) or shipping immature
SW.

Ralf