On Wed, 9 Jul 2008, Thorsten Leemhuis wrote:
On 09.07.2008 12:51, Panu Matilainen wrote:
At long last, we are about to get a brand new RPM version (alpha snapshot
at the moment) into rawhide. The list of changes from 4.4.2.x is massive
and a full summary needs a separate posting (will follow as time permits),
this is just a heads-up of immediate consequences for Fedora packagers and
rawhide consumers:
Sounds great, thanks Panu and others! Much appreciated and looked forward to.
But this announcement made me wondering: We have a big and complicated
Feature process [1] in Fedora that keeps a whole lot of people and committees
(especially FESCo) busy. Afaics the new RPM version is something that can be
considered a "feature" [2]. It was afaics not approved yet by FESCO [3] or
even proposed [4].
Hohum... fair point. Frankly, I've been so buried up in upstream rpm
development for a good part of the last year the thought of the feature
process never really so much as crossed my mind. My bad.
The new RPM is not committed yet or built yet, partly due to other issues,
partly to give time for any last regrets. Since we're obviously
having some last regrets here...
The simple-minded bass-player in me has just one practical question: what
do we do about it? Sure we can go through the feature process, it'll just
probably mean the new rpm will miss F10 alpha. Whether that's a good thing
or not might depend on if you're in rel-eng or not ;)
I would expect going backwards to an older RPM in rawhide
later will be next to impossible or very very hard. IOW: once it's in rawhide
for a few days FESCO kind of has no other chance then to approve this
feature, in case it ever comes up for a Feature vote in a FESCo meeting.
FWIW, having a safe back-out route is the very reason why it's going to be
built with bdb-4.5.20 instead of "whatever's latest" - that version
permits going back and forth between rpm 4.4.x and current one without any
complicated db conversion procedures.
So is the most of the Feature process (and especially FESCo's approval)
useless overhead? It looks to me that the answer tends to be "yes" as long as
big features like this can easily creep in without going through the
established approval process, as long as the feature gets added to rawhide
early enough in the devel cycle.
Just wondering. No, I really don't want to stop the new RPM; there are likely
other examples (say OpenOffice 3.0) in rawhide (but going backwards there as
hard as with RPM). But I'm more and more wondering if the complex Feature
process is worth all the trouble and effort. The best thing that came out of
it in F9 IMHO were the good release notes and great "whats new" pages. But
I'd say we can have that easier.
One reason for missing out on the feature process is probably that I've
found it somehow alien thing to begin with - I don't consider myself
working on a "Fedora feature", I'm "working on rpm.org upstream" to get a
much-needed update to the aging RPM version we have been living with for
ages. Mind you, this is not an excuse for missing out on distro policies.
The line between a "feature" and a "non-feature" is extremely obscure
really, and I think the point of "if you're unsure, ask" has not been made
sufficiently clear. Until perhaps now :)
- Panu -
--
fedora-devel-list mailing list
fedora-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/fedora-devel-list
