Le 21/06/2010 22:04, Chris Nighswonger a écrit :
On Mon, Jun 21, 2010 at 3:40 PM, Paul Poulain <paul.poulain@biblibre.com> wrote:
Le 21/06/2010 20:44, Chris Cormack a écrit :
This is certainly a possible plan of action, but no, I won't be pushing any patches to master without them going through QA. Certainly not 450 in one go. They will need to be branched into smaller feature sets and each one tested and merged. What I am willing to do, is be less strict about accompanying tests as the patches were written before that, what I am not willing to do is merge 450 patches that haven't been looked at by QA or the Release Manager into master. These aren't new rules.
1- we did a lot of QA on them, and although there may be some remaining bugs, we are live with them on 1 site, very soon on a second one, and many more.
But this approach is not acceptable in the community at large as evidenced by the BibLibre acquisitions work introducing four blockers.
I never wrote that publicly, but I must say i'm not very proud of the new acq code. it can/must be improved a lot.
What you and your clients may be able to live with, others may not be able to live with. So not only do we need good QA at the vendor level, but also at the community level as well.
of course. the idea being that merging very early (reminder : everything is written) means we would have 6 months to find & fix bugs. It's not the same as merging 10 days before releasing imo.
2- we proposed to submit those patches many months ago, but galen decision was : "feature freeze for 3.2". We had to go ahead.
This had to happen some time.
agreed. The problem being that 3.2 is very late. Today, we have to do something to merge our work.
I would suggest that the problem is not with the feature freeze, but perhaps with not keeping in sync with the main repo master.
we tried, but had some bugs introduced by the merge + we were too short on time. Again : today, we (BibLibre) are willing to do something because in a few weeks, it'll be too late, unfortunately
I think that at a minimum the present BibLIbre work would need to be merged into a topic branch off of the stable 3.2 master and then testing/debugging be done on that topic branch while keeping it in sync with the stable master.
see answer to the other chris for the rest of your mail ;-) -- Paul POULAIN http://www.biblibre.com Expert en Logiciels Libres pour l'info-doc Tel : (33) 4 91 81 35 08