Minutes Mar 10 2010
LSB Conference Call March 10, 2010
Stew Benedict (LF), Jeff Licquia (LF), Robert Schweikert (Novell), Kay Tate (Novell), Alan Clark (Novell), Alexey Khoroshilov (ISP/RAS), Jiri Dhulos (Novell)
LSB 4.1 status.
Jeff: LSB 4.1. Most of the spec work being done are 4.0 fixes.
This week been tied up with sdk refresh for 4.0. Some issues yet with these builds, plus a new issue with the way deprecation is handled in the headers. Will need to re-roll. Once rc is out, will be conservative about additional changes.
Putting together schedule for the F2F in April.
Jeff: Stew, should makelsbpkg stay at 4.0.90?
Stew: only at that version because it's pre-4.1 at the moment
Jeff: will look like it's pre-4.1 if we issue update
Stew: layout at the moment is kind of tied to the sdk version, may need to look at repackaging
Jeff: will work it out in IRC later
Jeff: dtk-manager -> dist-checker upgrade issue on Debian based distros. Probably a conversion issue we need to fix
Jeff: LSB 4.1
Much of the activity lately has been bug fixes that both feed back into 4.0 and 4.1. In the process we've uncovered more sdk issues.
As such, true 4.1 devel have taken a backseat.
Once the sdk gets out, we should be able to focus on 4.1
Pending: new cups symbols, symbols for a LSB compliant java build
How much can we get done before April? Freeze it then or need more time? t2c-alsa a factor also.
Stew: going to try and add packaging part for the minimal tests ISP/RAS added this week
Alan: what does the freeze mean without alsa tests?
Jeff: If there are no tests at all, we don't freeze. Maybe if we have some coverage (10%), we decide whether that's good enough.
Alan: Do we need to wait for the F2F to have that conversation?
Alan: What needs to be done for the alsa tests
Jeff: A package in snapshots and dist-checker knowing how to use it, plus results
Alan: Who's working on it?
Jeff: Stew, ISP/RAS, one other contributor
Alan: What portion has been done?
Stew: markup, no conversion to tests yet
Kay: Has the current markup been reviewed
Stew: Some, not all of it
Alan: Sounds like there is a lot of work to do yet, Alexey?
Alexey: Depends on the test quality required. There is a set in place now with minimal testing. "normal" quality tests require 1 man-day per 1-2 interfaces.
Alan: how many interfaces do we have, Stew?
Stew: several hundred I believe
Alexey: We can do some shallow, some normal
Jeff: So we have to make a qualitative judgement at some point whether it's "good enough". We basically knew this going in that we would not have full, good coverage with the man power we have. But we can improve on whatever we release in 4.1.
Alan: has anyone looked at the upstream tests?
Jeff: they're basically hardware tests, don't fit our model
Alan: aren't we going to hit some of these same issues trying to test the interfaces?
Stew: possibly, interfaces I've marked up so far look like they could be tested with minimal test envrionment
Jeff: we may need to choose to only test things we can do without accessing hardware
Alan: How do we get the alsa team to adopt what we put together and maintain them
Kay: Will alsa be that important a couple of years out, or will we be looking at a higher level API?
Jeff: We know there are higher level APIs gaining ground, pulseaudio beeing prominent at the moment
Jeff: Alsa still is lowest common denominator at the moment, I think we're safe in putting future effort into it.
Jeff: upstream is open to new tests, we need something to present to them
Alan: how does one patch/fix the current work? Do I use kompozer, submit a patch?
Jeff: make the new markup, then re-generate tests, is my understanding
Alan: So what is being maintained now? xml + test code?
Stew: Just the html spec files right now
Jeff: Markup is being done on copies of the upstream spec.
Alan: Not clear to me how it will be maintained
Alexey: Our approach has been to work from the LSB referenced frozen version. When LSB changes to a new version, there are several steps we have to go through to make sure the tests track the new specification.
Alan: Has this bee written up anywhere?
Alexey: Not yet. But a good idea
Alan: Can you write up the process and share?
Alexey: We will try to add something to the wiki
Jeff: Who is going to own the deliverables for this?
Stew: I'll do the packaging
Jeff: Stew and I can work out who is going to put the pieces in place to generate code from the markup
Jeff: closing remarks
End of Call