Plan for tomorrows (20070517) FESCO meeting

[Tom "spot" Callaway]

On Thu, 2007-05-17 at 11:18 -0400, Christopher Blizzard wrote:
> On Wed, 2007-05-16 at 17:36 -0700, Toshio Kuratomi wrote:
> > 
> > 
> > I think we need to discuss whether the Packaging Guidelines are only
> > Guidelines or if they're Rules.  If the latter, whether we're willing
> > to
> > take any actions to enforce them.  If turnout is light, I don't expect
> > us to make any decisions but we can't continue to debate whether it's
> > okay to "bypass the guidelines".  We have to know what standing they
> > have. 
> 
> Don't enable the rule police.  Guidelines are "best form" but I don't
> think they need to be enforced to the level where it becomes crazy.
> Remember, we are here to enable developers so we need to be reasonable
> people.  (An exception process?  Come. On.)
> 
> Honestly, I believe that this decision right here determines if we turn
> into structure for structure's sake or we turn into something that is
> open to developers.
> 
> We need to stick to the middle ground here.  Guidelines, not rules.

I understand what you're saying, but we've got maintainers saying "I
ignore the guidelines". This leads to bad packages.

All we're really trying to do is make good packages. We've tried really
hard to make guidelines that lead to good, clean,
maintainable-long-after-you-are-dead packages.

All we're asking (well, maybe all _I_ am asking) is that when people
feel like they need to break one of the guidelines, they present their
case and document that case in the spec file. I know that most of our
packagers are smart folks, but LOTS of people learn how to write rpms
from viewing spec file examples. When you're violating the guidelines
for a valid corner case, and you just do it, someone else comes along
and thinks its OK to violate the guidelines for their normal case.

IMHO, none of the following are valid rationales:

* "Always did it that way."
* "I felt like it."
* "Guidelines are dumb/silly/lame/broken."

These are valid rationales:

* "This package doesn't work without static libs." 
  (The packager may not be able to fix the code to not use static libs, 
   but by making the problem publicly known and internally documented,
   someone may come along and fix it for you. Yes, this has happened.)
* "My package is doing something unique that falls outside of the 
   guidelines."
  (When we see this, we make an exception for your package, and update 
   the guidelines so that it is covered. If you never report it, we 
   never know.)

The Fedora Packaging Guidelines are living documents. Anyone with wiki
access can submit a draft to add/change/alter/bend/mutilate them. The
process is documented here:

http://fedoraproject.org/wiki/Packaging/Committee#GuidelineChangeProcedure

Thanks,

~spot