* Ayal Baron <aba...@redhat.com> [2012-10-02 18:41]:
Thanks for taking time to write up your thoughts.
> > >> A polite and well-detailed -1 should be used daily and not
> > >> considered
> > >> "rude".
> > >>
> > >> Dan.
> > >>
> > >
> > > Yes, this is the bookmark approach...
> > >
> > > The problem is that most people has the -1, 0, +1 scale, and in
> > > that scale -1 is actually the reject method.
> > >
> > > "If I was the maintainer I would have rejected this patch".
> > we could open -2 to non-maintainers as well if it is of use.
> > danken?
> Personally I disagree with this approach, -2 should rarely be given
> and doesn't disappear. It should only be given to people who really
> understand the code properly and whose opinion is well accepted by the
> I agree with Alon that 0 can and should be used, I do not agree with
> him though that ack/nack type of discussion should take place on it
> though and I'll explain why -
> First of all, in gerrit there is no immediately visible difference
> between '0' and no review at all so someone might have serious issues
> with a patch but if she did not mark it with -1 submitter might
> totally miss this fact. esp. if someone sent a new revision and the
> title of the cover comment for previous version doesn't state a -1 (so
> maintainer doesn't know he needs to go looking back to verify
> things were fixed).
Absolutely. I've been worried about this specific issue. For me, at
least, the gui tends to hide quite a bit of the conversation in a normal
discussion of a patch. The comments section in the gui many times only
says '1 inline comment' and provides no direct link to the text of the
comment. Following the thread via email is much easier since all of the
comments are there and ordered, threaded. However, we currently can't
reply/participate via email and also include the comments back into the
gui for the other users.
So I've been concerned that conversations might be lost as new revisions
of the patches come in.
> This is adding overhead on maintainer now to go back to each and every
> review and make sure that there are no comments that should have been
> addressed in 0. Note that if someone gave a -1, normally I'd expect
> that person to make sure and +1 a subsequent patch to flag to
> maintainer that all their problems with the patch have been addressed.
In addition to maintainers, any community member attempting to help
with the review load, this is an issue as well.
Software Engineer; Linux Technology Center
IBM Corp., Austin, Tx
vdsm-devel mailing list