svn commit: r287780 - in head: share/man/man9 sys/kern sys/sys

[hiren panchasara]

Couldn't have said this any better.

On 09/19/15 at 06:38P, Adrian Chadd wrote:
> What isn't necessarily public knowledge is the sheer volume of emails
> that went out a few months ago whilst chasing down callout and tcp
> bugs. There were (and maybe still are) very subtle bugs in the callout
> system and after a few attempts at fixing them there were some very
> careful bug fixes made. Some attempts failed, I think a couple of
> successful ones made it into the tree.
jch@ and hans have been working on callout lately. Both collaborate on
reviews/commits, take each other's suggestions and catch each other's
mistakes. In this particular instance, rrs@ didn't like the change and
he asked some questions. Hans responded to that. Which is pretty normal
for this list. What I do not understand is, why is everyone coming out
with "please revert right away"??

> 
> Yes, this whole callout system is very delicate at the moment. hps@
> has some very specific ideas of how the API should behave in order to
> be predictable/reasoning-able (and I agree with him about almost all
> of it, even though it makes RSS painful, but that's because of our TCP
> stack and how we use callouts, not because its his fault!) but it's a
> pretty big fundamental change to how things currently work and he was
> shot down. I think people are just very weary of new changes.
> 
> On the flip side, he did actively solicit reviews - rrs, kib, hiren,
> jhb, wblock and jch were included in the review request, which dates
> back to August 28. He gave people a little short of three weeks for
> review before he committed the code. So as much as I'm cautious about
> things (and it gets me in trouble at work, hi alfred!) I think he did
> the right thing here - he added a new thing, documented it, solicited
> a review, and it timed out. If people would like more time to review
> it then fine, but please give him either a firm "no, not ever" right
> now and be honest about your intentions, or give him a timeframe that
> you'll review it before it times out.
In principle, jch@ agreed to the review/change in question so it'd be
incorrect to say that hans made this changes without anyone's knowledge.
Now, without any other reviewers commenting on the review or asking him
to wait for the review for 3 weeks, how long should he have waited
before committing the changes?

> 
> Hans - personally, I think you should've emailed out a review request
> on freebsd-arch@ and put out a request for testers and give a firm
> date that you'll commit it. That makes it all very explicit.
> 
That indeed would have made things clearer.

Hans - if you think that questions raised by Randall would take more
iterations to get answered, please revert the change and discuss it on
the review you already have opened for this issue. Also, try and get
explicit YES/APPROVED on callout related reviews from jch, rrs and
others. (Others: please speak up so hans can add you to this review and
future reviews.)

> People channel phrases involving silence and agreement and all that;
> this is one of those times it happened.

Cheers,
Hiren
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 603 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/svn-src-head/attachments/20150919/f9c0c331/attachment.bin>