ESR "fedora-submit"

[Eric S. Raymond]

Warren Togami <wtogami at redhat.com>:
> >Not only did you back the idea, you came fairly close to writing a
> >spec for it.  I don't know what side of the bed you got up on this
> >morning, but I suggest you try the other one.
> 
> This was a *LONG* time ago, long before the current Fedora package 
> process.  Subsequent to me writing those suggestions, your idea of 
> fedora-submit and desired method of software submission turned out to be 
> missing the point entirely.

Let's suppose I had a package approved for Extras, or whatever Extras
is called this week.  Would it still be necessary for me to do work by
hand to get a point release into the repo, as opposed to being able to
script the process?

If your answer is "yes" then, from my POV as a maintainer of thirty-six 
projects, the submission system is still broken as designed.  It imposes
an overhead on me that I find unacceptable.

> (I am unable to find the post in the mail archive now, but I vaguely 
> recall you wanting a "contrib" style upload system where you could 
> submit RPMS for Fedora to directly consume.  If my memory has properly 
> attributed this to your desires for Fedora, then this was furthermore 
> crack.)

Your memory is faulty.  I specifically told you (and I can produce the
email to prove it -- I reread it less than an hour ago) that I *don't
care* whether the update packet I'd have to ship is an RPM, a tarball,
or a patch and a job card.  

The functional point is that *it must be possible for me to fully
script my release process*.  I'm willing to do hand-work on initial
submissions, but not on update releases.  If I can't fire-and-forget
those, I won't play.  Dammit, I don't have the *time* to jump through
clicky-webby hoops or go through protocol exchanges I have to pay
attention to with my one precious brain.  Not over thirty-six projects.

Your apparent inability to grasp this point, instead obsessing about
technical details such as whether or not the submission queue is "RPMS
to be consumed" speaks to me of a fundamental failure in Fedora's
vision of its relationship with developers.  You're not seeing the
forest for the trees.  You're passing tactics but failing strategy.

> Fedora has a review process that has allowed the project to scale in a 
> responsible manner.  Dozens of Reviewer/contributors have been promoted 
> as they did work over time.  THOUSANDS of packages have been approved. 
> The backlog of the review queue has shrank to a manageable level, 
> indicating that the process has been working.

*sigh*  Fine, fine.  Motherhood, apple pie, wave the flag.  I'm happy
for you that you think your process is working.

The question remains.  Do I have to *do shit by hand* to get point
releases into the receiving end of your wonderful review process?  

If I do, your design is still unusable for me.  No matter how
marvellous you think it is, it fails to scale to my needs.  

I can't stop you from dismissing this evaluation as 'on crack', but I
can and will simply dismiss you as an oblivious idiot if you do. 
 
> The current incarnation of this process uses Bugzilla during review, and 
> CVS there-after for package updating by the maintainer.  This process 
> has done a tremendous job of growing Fedora's software catalog.  Further 
> improvement is on-going, with better tools and processes to reduce 
> maintainer and process overhead.

Can I script an update to the CVS?

Can I write a procedure in shell or Python, to be invoked by 
'make ship' from my project's top-level makefile, that will automate
my release process so I *don't have to think about it*?

I *don't* *fucking* *care*  what happens on the back end. If one
of the possible consequences is email bounceback that says "your
point release failed QA because *BLAH*, that's OK.  I'll fix the
problem and try again.

> In any case, please remove statements from your page like, "It will ship 
> with a future release of Fedora Core." and my explicit endorsement of 
> your tool, because they are simply not true.

Would you prefer "Warren Togami liked this idea when I proposed it,
but has since gone as crazy as a rabid dingo and changed his mind?"
Because that's how reality looks to me right now.
 
> You claim that Fedora's submission process even today is a failure.  To 
> date you have never even TRIED to submit a package to Fedora or to 
> maintain a package here.

That's right. If it isn't obvious to you why I haven't by now, you
really are insane.  It's because you made the perceived overhead 
of submitting a package and update releases too high.  

Your submission process scared me away.  *Me*.  Mr. Open Source
himself, as experienced and battle-scarred a developer in this
mode as walks the planet and breathes.

That makes your submission process a failure.

I've still got the bloody printed application form on my desk.  But
I never filled it out because nobody persuaded me that I wouldn't
be letting myself in for a nightmare of bureaucracy and repetitive
hand-work.

Go ahead. Persuade me.  I would love to believe that all the sweat 
I've put into improving RedHat/Fedora, and all the abuse I've
taken from my Debian- and Ubuntu-loving friends, was not in vain.

My previous offer still stands.  If you're willing to make it 
technically possible for me to automate pushing my point releases
into your queue. I'll write the client tool to do it.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>