geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rodent of Unusual Size <Ken.C...@Golux.Com>
Subject Re: Change to commit model for Apache Geronimo
Date Wed, 14 Jun 2006 22:14:59 GMT
Hash: SHA1

John Sisson wrote:
> I agree that "merging" shouldn't require another RTC.  So merging of 
> your m2 migration changes should be OK.

It's not necessarily as clear-cut as that.  Any change
that's going into something provided as part of an
Apache release needs to go through the approval model.
Hack away in the sandboxes freely, but rolling anything
from a sandbox into a intend-to-ship line needs to go
through the model -- RTC, in this case.

Merging in changes from another branch can be a bit grey,
as well.  Why weren't they in the target branch?  Did they
have anything to do with why the source branch was
abandoned (in this case)?  (Those are rhetorical questions.)

It's a judgement call on the part of the committers.
Since RTC is intended to emphasise quality over development
speed and convenience, the question that needs to be
asked and answered is along the lines of, 'Is this a
functional change, and is there *any* danger of it
introducing bugs that would be caught by reviewing it

Committers have to be honest with themselves and the
project, and answer the question objectively rather
than taking an easy way out.  'Fifty-plus files is
'way too much to review, so let's just assume that
since it was in the other branch it's okey to bring
in here without checking' is a *wrong* answer.
'There are fifty-plus files involved, but the changes
were all exercised and tested in the source branch, and
they're being merged into code in the destination branch
that is the same as in the source branch, so I feel
confident merging it won't tickle any bugs from the
integration, and I don't think we need to review' is a
valid answer allowing unreviewed merging.

Cases like this are up to the committers to decide.
Based on that definition, if no-one honestly thinks the
merge could introduce bugs, then commit away.  Remember,
though, that even in CTR someone can discover an
unanticipated problem and issue a veto..
- --
#ken	P-)}

Ken Coar, Sanagendamgagwedweinini  http://Ken.Coar.Org/
Author, developer, opinionist      http://Apache-Server.Com/

"Millennium hand and shrimp!"
Version: GnuPG v1.2.4 (MingW32)
Comment: Using GnuPG with Mozilla -


View raw message