On 6/15/2016 3:35 AM, Paul Jakma wrote:
On Tue, 14 Jun 2016, Jafar Al-Gharaibeh wrote:
I see it happens a lot where the same fix is submitted over and over
again by different developers! we could use their time in better ways :)
Clearing backlogs, and automating some integration work will go a long
way to speeding things up. There's still a good bit of work to do on
the backlog, but the end is hopefull in sight and things might go more
smoothly there-after.
Bug fixes should go into master within few days at the latest, no
need to hold them off. Maybe the first ACK should trigger such
patches to be queued up to go into master as soon as one of the
maintainers get a chance to do it.
Can you define 'bug fix'?
Yes, a patch that fixes a crash or a wrong behavior or... ;)
There are situations when a bug fix might be very big patch or discussed
for an extended period of time, but in most cases as I mentioned above,
we have really trivial bugs, null pointer deference, off-by one errors,
initialized with a wrong value etc. There is really a great benefit in
having those in as soon as possible.
I'd love a way to be able to reliably and quickly sort contributions
into 'obvious quick bug fix' and 'other', but - and I've been doing
this a while - experience is that isn't always so easy. E.g., "big"
feature-add patches can sometimes be easier to get in than little 'bug
fixes', if the feature add is well-contained and doesn't affect other
code and the bug fix raises architectural issues.
Get the obvious ones quickly then and leave the non obvious for the
"slow" process/review.
As I'm fairly sure now that the required magic wand to fast-path
'obvious bug fixes' doesn't exist, I think the only way left is
instead to sift out the hard stuff and get it out of the way of
blocking other stuff. So everything goes down a common initial path,
unless someone gives it a 'kick' onto a slower path for whatever
reason. Which implies that speed requires optimising that common path.
I wasn't expecting a magic wand that can fix and sort out things now,
I'm suggesting this as a future option when all of the current backlog
is out of the way.
Regards,
Jafar
regards,
_______________________________________________
Quagga-dev mailing list
[email protected]
https://lists.quagga.net/mailman/listinfo/quagga-dev
