BIND 10 Team Call

2011-07-19 15:00 UTC



Review of msgq replacement options

Main options seem to be dbus or MPI:

Stephen: A comparison (grid) would be really nice.

Shane: I guess as a starting point.

Stephen: Do we have a list of what we can't do without?

Shane: Anything not "nice" in ticket #765.

Jinmei: Can we make the decision in the face to face meeting?

Shane: I guess there is no reson not to. No specific deadline for this.

Stephen: It would be easier.

Shane: Still need the comparison chart. Will have discussion on Tuesday.

Mid-Sprint Defects

Jinmei: Build problems with old Solaris systems on Jabber.

Larissa: In BIND 9 we have been working on guidelines for bugs to add to a sprint in the middle. It might be good for BIND 10 too.

Stephen: Is this for bugs that come up because of changes, or ones that were already there?

Shane: I think it means both.

Stephen: For bugs committed in this sprint should be fixed in this sprint.

Shane: At some point a bug is so serious that we have to include it. Only very serious bugs like a security issue?

Stephen: I think so.

Jinmei: "Sprint is full" cannot be a reason to exclude bug fixes. Sprint will always be full!

Jeremy: Maybe we should re-order the call to discuss defects for the first 15 minutes or so.

BIND 10 schedule shift

Reminder: Shift to BIND 10 schedule due to 3-week sprint.

No sprint planning... do we want to have a team call next week?

Jinmei: Not so many people will attend the IETF, so it will make sense to have a call at the (bi)weekly one as usual.

Shane: I agree. We'll do a call just like this.

Jinmei: No reason not to have daily calls.

Shane: I agree. AP: send mail, adjust calendars, etc.

Jeremy: Should we make the calls all start at the same time?

Shane: It's not nice for Jinmei and Larissa...

Post-release Discussion?

Jeremy: When should we do that? 2 weeks after? Is that too late?

Shane: Should we move into the sprint planning call?

Jeremy: In general we don't have time.

Shane: Important for the whole team to have the discussion.

Larissa: It's better. But a good question.

Stephen: How soon after the release do we need to have this discussion?

Larissa: Very soon after. What if we added 15 minutes on to a daily call?

Jinmei: Another possibility is to release beforehand. Not having 1-week period before... have 6 or 5 days.

Jeremy: I thought about that. Friday snapshot releases are fine with me.

Shane: Is that actually enough time?

Jeremy: As long as we don't push a lot of code after Tuesday.

Shane: Okay we will try Friday release, and Tuesday discussion.

Dependency Management (from Jinmei)

Jinmei: This is about the message sent about Debian experience.

Jinmei: Is building everything a huge overhead?

Jinmei: One traditional solution is to make a Wiki page for external libraries per operating system / distribution.

Shane: I think that's reasonable.

Jinmei: Maybe we also provide another script in addition to ./configure which checks all dependencies and then shows the URL for each system.

Shane: A version of ./configure which does not break until it gets to the end? That would have been helpful.

Jeremy: We also talked about having a script which downloaded for you and installed. We decided not to.

Jeremy: We should also work on getting these dependencies put into package systems that are missing them. I've been doing that.

Requirement for update frequency (wrt multi core support) (from Jinmei)

Jinmei: Part of discussion for multi-core support for server. How often we should allow updates. 10-20 per second? 1000? How instantly we should make changes visible.

Shane: People want thousands... quickly.

Jeremy: In support we talked about that from cable providers. 10's of thousands of DDNS updates.

Stephen: Is there a difference there between when the updates are handled and when they are visible?

Shane: I think people expect to see changes "soon", but administrators may use dig to check

Jinmei: It may affect our approaches to handle multi-core significantly. We will need some specific requirements from expected users.

Stephen: Is this something we could ask on OARC?

Shane: We can, but there may be missing big DDNS users. Also some people may not use DDNS at all because BIND 9 is too slow.

Stephen: We need a decision soon.

Jinmei: Product manager or someone with a connection to the user is needed.

Larissa: We can do that.

Shane: How urgently is this needed?

Stephen: How soon do we want the refactoring to begin.

Shane: Larissa and I will talk about this.

Whether to support multiple questions (from Jinmei)

(may be better to discuss this with more attendees)


Last modified 7 years ago Last modified on Jul 21, 2011, 10:48:42 AM