FESCo: Feature process and release blocker process

Adam Williamson awilliam at redhat.com
Fri Jul 15 02:36:10 UTC 2011

As FESCo doesn't have its own mailing list, and this isn't a trac item
exactly as I'm not really requesting any *action* on FESCo's part, I'm
sending this to devel and test in the hopes it reaches all parties
concerned. This relates mainly to the feature process and hence is of
particular interest to FESCo (which owns the feature process) and Robyn
(as the feature wrangler).

There have been a few occasions in recent releases in which bugs that
can essentially be characterized as 'the proposed feature XX is not
complete' have been marked as release blockers. When these have come up
for review as part of the release blocker process -
https://fedoraproject.org/wiki/QA:SOP_blocker_bug_process - we have been
consistent in not accepting them as blocker bugs.

Our rationale for this builds from these premises:

* The release blocker process is designed to ensure a given release or
pre-release's compliance with the Fedora quality standards, as codified
in the release criteria -

* A feature being incomplete does not necessarily, or even commonly,
constitute an infringement of these quality standards

We - QA - would like to formalize this position by writing it into the
blocker bug process / release criteria (I haven't yet looked at
precisely where it'd fit best). We feel that it's best to keep the
release blocker process and the feature process separate.

In case anyone's worried, let me clarify that this is entirely a process
issue: its practical consequences are small. The idea is not to mix
together the feature process and the release blocker process. It would
still be the case that FESCo could, as part of the feature process,
choose to delay a Fedora release or pre-release if it felt that a late
proposed feature was important or delicate. QA is not the only group
that signs off on Fedora releases; if there are no blocker bugs, QA will
say that the release meets quality standards and QA is happy to sign off
on it, but this does not preclude FESCo from deciding not to sign off on
the release on the basis of an important or delicate feature being
incomplete. We'd just like to make it clear that such a decision would
be made by FESCo as part of the feature process and tracked however
FESCo liked to do it, rather than doing it by marking a bug relating to
that feature as a blocker, hence mixing up the feature and blocker bug

We wanted to check that this was okay with FESCo and the feature
wrangler and the project in general before going ahead, so here we are
=) Please let us know if anyone is worried about this. Thanks!
Adam Williamson
Fedora QA Community Monkey
IRC: adamw

More information about the devel mailing list