Hi,
The core process you describe isn't substantially different to what
we've been developing since last July:
--------------------------------------------------------------------------
1 Systematically collect all patches since the last integration round,
and linearise them (i.e. queue them up, work out any conflicts if
possible - push-back anything that can't be reconciled to the
contributor) into a labelled 'round/$X/proposed' HEAD in git.
2 Publish the HEAD and a summary of what's in it to the list, and ask
for any final reviews, with a bounded period.
3. Based on the reviews, sort the proposed patches into 'accepted',
'deferred' and 'rejected' heads (the latter probably better called
'pushback').
4a. Have a final call on the 'accepted' head.
4b. (More recent addition) NetDEF do basic protocol testing (in addition
to the automated, per-contribution basic CI build tests they do)
5. Merge the 'accepted' head into the main 'master' of Quagga
Rinse, lather, repeat. Ideally on a bounded basis (aiming for monthly).
-----------------------------------------------------------------------
The above process can be optimised. E.g. there are parts still to be
pipe-lined better; there are parts that can be automated; there are
parts that can be distributed.
I'm fairly confident it can deal with things even as is. Hell, I've
already got most of the outstanding patchwork patches queued up. The
answer to our problems is just to knuckle down and get stuff done.
Quagga in 2015 saw more commits than any year since 2004, inc the pimd
merge (~170 commits at most) or 2006 excluding:
By year
2003 101:*************
2004 554:*********************************************************************
2005 457:*********************************************************
2006 239:******************************
2007 126:****************
2008 105:**************
2009 185:************************
2010 59:********
2011 340:*******************************************
2012 357:*********************************************
2013 134:*****************
2014 133:*****************
2015 478:************************************************************
2016 76:**********
Here's the monthlies, since we started rounds, which compares fairly
well to monthly averages for years above (201502 is mostly pimd):
201501 17:*******
201502 170:*********************************************************************
201503 9:****
201504 51:*********************
201505 40:*****************
201506 31:*************
201507 10:*****
201508 9:****
201509 47:********************
201510 42:******************
201511 1:*
201512 51:*********************
201601 0:
201602 56:***********************
201603 20:*********
201604 0:
201605 0:
I don't know why there's been nothing the last few months, but I'm
preparing round 8 'proposed' now, which should pretty much get us caught
up.
The stuff about releases and what not is a distraction. People are
welcome to go tend stable releases. Different flavours of 'stable' are
possible - don't need consensus on that. Personally, I think that's
stepping on the toes of vendors, and not per se Quagga.net can do better
- but I wouldn't stop someone from trying. I'd suggest separating that
discussion to make it easier for people interested in that.
On the Google document, there's at least two things there you know I'm
not at all comfortable with, so I'm just going to stay away from it.
regards,
--
Paul Jakma [email protected] @pjakma Key ID: 64A2FF6A
Fortune:
Small change can often be found under seat cushions.
-- One of Lazarus Long's most penetrating insights
_______________________________________________
Quagga-dev mailing list
[email protected]
https://lists.quagga.net/mailman/listinfo/quagga-dev
