Minutes Jun 8 2011
Jeff Licquia (LF), Robert Schweikert (Suse), Mats Wichmann (Intel), Stew Benedict (LF), Alan Clark (Suse), Russ Herrold (Owl River)
- LSB 5.0 planning and tasklist.
5.0 Project Plan (wiki): http://www.linuxfoundation.org/en/ProjectPlan50
Jeff: Started the thread on time-based release on mail list, continue it there. Need to get our priorities right on the task list.
Jeff: Are the tasks on the list correct? Things missing or there that shouldn't be?
Mats: gcc-4.5 support should also include 4.6, which is stricter. libstdc++ uplift ties into this. Going to get progressively harder to use old libstdc++.
Russ: There are other compilers (clang/llvm)
Mats: there are, but gcc is probably the one the most people will have available
Russ: May fix gcc strictness issues, since clang is more strict than gcc in general
Robert: do we care about the compiler beyond what it produces in ELF?
Mats: SDK issue, more than a spec issue
Jeff: Are all these tasks specific 5.0 goals?
Mats: Some could be version independent sdk issues.
Robert: We need to go through each item in the list, not generalize. Need to look at each library in the spec for possible uplift.
Jeff: So we need to create a list of each library status. Our version vs distro, vs plans going forward
walk down the list -
1) app-checker cleanup - needs a clearer goal, a bullet list of things to fix?
2) init - new competing init systems - too early for us to worry about, sysv init scripts will still be around for a while
Mats: not that clean, configurable in systemd, some people are turning sysv compat off.
Jeff: maybe we need tests for init
Stew: We have some, install_initd, etc. lsb-apache installs an init script and the FVT checks that it's enabled
Robert: Need to re-visit in a year or so. Possibly deprecate SYSV.
3) ALSA-T2C - need to improve test coverage, specifics?
Stew: Have been working on adding some more conversions from shallow to normal testing based on Navigator usage stats (Another 20-30 interfaces).
4) SANE - need to reach out to printing group for domain expertise. Might be able to put some resources into it.
Robert: Don't see SANE as a blocker to 5.0.
Mats: Printer driver people want it, not necessarily tied to a particular LSB version
5) Version uplifts - need to look at all the libs
6) openGL uplift
7) libstdc++ uplift
8) DBUS - high priority, needs to happen, need to workout timing
9) gcc-4.5 support - see discussion above
10) modularization -
11) test framework - how to work tests back into upstream? is this possible?
Russ: test advocacy might be done best by just leading, it's not as hard as it used to be to run the LSB tests
Robert: the LSB tests are basically unit tests, which we have to maintain. We find the problems too late, when it's already in the distributions.
Jeff: Many of our tests are ported upstream unit tests. Results of which are mixed. Is this goal realistic, does it buy us anything?
End Of Call