The Linux Foundation

 
Minutes May 25 2011

From The Linux Foundation

Attendees

Stew Benedict (LF), Robert Schweikert (Suse), Jeff Licquia (LF), Jiri Dluhos (Suse), Russ Herrold (Owl River)

Agenda:

  • FHS status.
  • LSB 4.0 refresh status.
  • How to bootstrap a new architecture? (ARM, if you're wondering.)

Minutes:

Jeff: FHS status, got bugs filed and rollup put together for 3.0. Posted to list for comments. Non-controversial things, things already in distros, typos, build issues are what's in the list now. May start into the build issues soon.
Jeff: Participation has been good on fhs-discuss, a bit heated at times, but active.
Jeff: Comments?
Jeff: 4.0 refresh continues to move forward. A few issues have come up, xts5, appbat. Have come out of testing.
Jeff: Appbat build has "check" rules. Wasn't backported to 4.0, check rules are added to 4.0 now.
Stew: Intrisic issue with the way we build appbat, we set PKG_CONFIG_{PATH,LIBDIR} to /opt/lsb/lib{,64} and once the SDK moves on things (libbat) are not longer there, we have to point to the specific version (lib-4.0).
Jeff: doing the fix in 4.0 is probably the right thing to do, perhaps we need to fix the problem correctly in 4.1. Is there a bug filed?
Stew: Not yet, but I will.
Jeff: Anything else of the refresh? Stew you've been doing the testing
Stew: I think we'll need to re-test appbat once I get things to build
Jeff: Had a new person show up last week, had some discussion on ARM. Have people interested and willing to volunteer resources. We need to capitalize on this. How do we do this?
Jeff: Core LSB peple will have to do some work to give people a framework to work with.
Jeff: New architectures don't come along very often. New people are almost as rare.
Jeff: Thoughts?
Robert: My concerns are 1)which ARM architecture? 2)Do we have distributions that are interested? There will be maintenance involved after the initial work, and it's unlikely the people willing to do the initial work will want to maintain it.
Mats (not on call): yes, people from Fedora, Debian, Ubuntu have expressed interest
Jeff: As you said Ubuntu is doing ARM, and I believe Fedora is interested
Jeff: We have a couple of other architectures already in that mode, needing maintenance and not a lot of interest, what's one more?
Robert: Another question, where do we get hardware for testing?
Jeff: Good question. We'll have to see if someone has hardware to contribute.
Russ: I know Fedora has a large build farm, from GlobalScale. Possibly we could get them to do testing for us, or talk to GlobalScale
Jeff: Actual hardware or a cloud thing?
Russ: Hardware (See http://paulfedora.wordpress.com/2010/07/09/fedora-arm-buildsystem-status-update/)
Jeff: Have had some issues with hosted build platforms, but mostly workable
Jeff: How do we get started? As Robert mentioned, will take some of the core team's time getting people up to speed. Need database work, familiarity with the tools, etc.
Jeff: Mats and I talked about setting up an ARM fork, just a copy, something similar to ARM we copy, then fix it.
Russ: There are distros in the pipeline; Debian essentially exists presently; Fedora at the RHEL 6 equivalent level is close. Mostly a documentation issue.
Jeff: Does the above approach sound like a way forward?
Robert: Sounds reasonable to me, may be a benefit in seeing ways to make contributions easier.
Robert: Will distros actually support ARM? Will Red Hat support RHEL on ARM?
Russ: Community distros will
Jiri: (missed the comment)
Robert: Will what we create now, be applicable when Enterprise distros actually support it?
Jeff: I think it's worth pursuing
Jeff: I think the next step is to followup on the mail list and propose this
Robert: Need to solve the hardware question first
Jeff: Yes, post the problems and proposal
Jeff: Anything else on ARM? Any other architectures?
Russ: The mention of enterprise distros supporting ARM brings up the question of whether we should drop some, like IA64
Jeff: In terms of architecure pain (for LSB group), IA64 is currently one of the lowest. I usually use that builder first for build issue questions. We've had numerous issues with s390/x.
Robert: For IA64, the writing is on the wall. If we were to deprecate IA64 for 5.0, we would simply be following the trend. Can't get away with that for s390.
Russ: No reason to drop s390, enterprise usage is still prevalent.
Robert: Right, s390 isn't going to go away. It's a pain to support, but... we need to come up with a better way to support it perhaps.
Jeff: There are 2 architectures, the 31bit is less prevalent. Possibly the same for ppc32. Any time there is a 64bit version, you have to question the need to continue support the 31/32bit platform.
Russ: 32bit is still mostly all that's needed in virtualization environments
Jeff: Right, in the x86 world, but what about s390 and ppc32?
Russ: Doesn't IBM support it?
Jeff: Linux support is lagging.
Jeff: Anything else?
Robert: We need to get back to LSB-5.0. Other work is important, but we need to focus on this.
Jeff: Let's make that the primary agenda item next week.
Russ: FHS governance shouldn't be forgotten, needs to carry over until a draft is generated.
Jeff: I may start a thread on that in fhs-discuss.

End Of Call


[Article] [Discussion] [View source] [History]