[lsb-discuss] LSB conf call notes for 2008-01-16

[Jeff Licquia]

Attendees: Jeff, Dan Kohn, Sam Hart, Stew, Darren Davis, Marc Miller, 
Mats, Robert Schweikert, Marvin, George Kraft, Kay Tate, Ted Ts'o, 
Vladimir, Alexey

LSB 3.2.  Intro.  Sam: lots of info is redundant with data in the 
autotester.  Some tests are difficult to determine what failing tests 
are doing.  Will be posting info on IRC.  Stew: whittled down Perl 
issues, about half-a-dozen left.  Sam: on TestPilot32?  Stew: no. 
Autotest results are on http://www.linux-foundation.org/lsbtest/. 
Robert: shouldn't the deps be in the LSB?  Jeff: new stuff for 3.2 isn't 
in the lsb dependencies yet.  Stew: also, need more than the LSB to test 
the LSB.  Jeff: timeframe?  Sam: weeks with the current TestPilot32 
data; maybe should cap or focus on the autotester.  Maybe a few days if 
we limit to failures on several distros.  Stew: freedesktop may also be 
an issue.  Don't normally run because of the interaction issue.  Jeff: 
need to make sure they don't get dropped; need to be mentioned in DTK 
Manager.

Jeff: any bugs we know about that are serious/difficult?  Stew: doesn't 
look like it.  Core is looking a lot better.  Stew: pushed that change, 
seems like the long double problems are behind us.

Ted: why are we just finding these now?  Jeff: lots of data, getting to 
the rest of the data.  Stew: also, not doing wide testing until 
recently.  Ted: need to make sure we don't do this in the future.  Mats: 
the steps have been taken, but too late to help.  Ted: wiki for the 
autotester?  Sam: part of the problem is that things were finished 
during the holidays.

Ted: so keep working and discuss on the bug call?  Are we sure we can 
get this data evaluated by then?  Sam: the majority of the multi-distro 
issues, yes.  Ted: documenting

Mats: spec is basically ready, except for a C++ question.  Building; 
will push.

C++ issue.  Ted: will this cause us issues?  Jeff: don't think so. 
Alexey: suitable to postpone the spec part.  Only concern: getting it 
evaluated immediately after 3.2.

Sam: 3.2 sample implementation.  Lots of work to do; will that be a drag 
on resources for 4.0?  Ted: what issues?  Sam: new libraries, build 
quirks.  Link on status page.  Lots of little issues to overcome.  Kay: 
need to clean up for 4.0?  Ted: yes, assumed that the 3.2 SI wasn't too 
bad, but perhaps that wasn't the case.  Sam: also, more useful; virtual 
machines.  Current SI is mostly unmaintained.  Concern is to take time 
away from 4.0.  Ted: first step is tarball chroot, next step is making 
it more useful in other ways (UML, virtual, etc.).  Can we compile it 
once?  Sam: multiple archs is also an issue.  Also, some LFS changes are 
biting us.  Ted: can we scale back the SI; archs?  Sam: maybe rejigger 
the 3.1 SI.  Kay: the SI is required for certification.  Ted: need to 
figure out what is required.  Sam: can't even build 3.1; once that's 
done, shouldn't take more than a month after building 3.1.  Need to try 
on an older distro.  Stew: builds on SLES 10.  Sam: will try that.