The LSB Workgroup will be meeting Mar 1-3 2005 in Austin, TX. This page will hold the agenda as it evolves, and the meeting notes afterwards.
- Tuesday 9am (CST): introductions, setup, agenda review, etc.
- Tuesday 10am: regularly scheduled lsb-test confcall. Several lsb-test members will be at the meeting; for the rest it will give added visibility into a key LSB 3 deliverable. Two notes on this:
- will be on lsb-wg number, not the lsb-test number
- 2. this is one hour later than the regular lsb-test confcall time
- ISO DIS 23360 status, POSIX update, POSIX/LSB reconciliation (ongoing)
- packaging gap analysis
- update test development plan for LSB 3 - deliverables, schedule
- review C++ status: have all crosschecks passed?
- detailed bug review for LSB 3 - make sure blockers are addressed, make sure other bugs were correctly triaged as blocker/non-blocker
- review/update lsb-futures items / todo list, build priority list for LSB 4
- library gap analysis
- discuss workgroup outreach to upstream projects
- review LSB standing documents, esp. policy document. Need to adjust some "policies", Mats requests separation of policy and tech decisions into separate standing documents
- training for new LSB developers (if needed)
- future F2F schedule: Sun and Intel have offered to host future meetings, there are also conference co-locate possibilities (OLS, LWE, and others). Co-locate probably needs to be done as a miniconf before an event, there's not enough time during an event for those who would attend. Meeting locations should not be limited to North America if we continue the quarterly meeting frequency which roughly describes the last three meetings (Aug-Dec-Mar). The ISO Ballot Resolution meeting is to be held the week of Aug 29-Sep 2 in Singapore; the host has agreed that at least part of this week could also be a LSB WG meeting (there is some dependency on the number of ballot comments received).
Andrew Josey (phone), Nick Stoughton, Matt Taggart, Marvin Heffler, Rajesh Banginwar, Kevin Caunt, Gordon !McFadden, Stuart Anderson, Mats Wichmann, Andrew Cagney, Arthur Tyde, Rita Gupta (phone), Nilesh Jain (phone)
Reviewed items on Test list at ProjectPlan3, that list has been updated.
- AI: set up centrally maintained waiver file (extraction from cert db?) that tjreport can access. tjreport already knows about database access and waivers files. Note this sort of thing has to be multi-arch, multi-version.
Discussion of adding command tests, some existing ones use expect, if need to rely on a specific version of expect can use LSB appbat tcl/expect. Some commands already come from usersgroups and i18n testing.
- AI: command tests - report on which commands are completely untested
Checker tools: several of the todos are done (so marked on ProjectPlan3). archk work can maybe get done as part of existing work plan for appchk. dynchk is now working, need to move forward. More typetype descriptions are needed to keep making it more useful, but we need to get it out to stimulate usage. Dynchk only works on 2.6 kernels for now. Is there a resource to run the appbat FVT checks under control of dynchk?
- AI: turn on dynchck autobuild
ISO / POSIX report
- AI: look into usability of
__xpg_strerror_r from glibc_2.3.4
- AI: Mats to add wording and make proposal on certification timeout. Intent is to allow two major certifications to be active concurrently, but not more.
Workgroup to more actively maintain the document, standing item at workgroup F2F and calls.
In discussing the need and schedule for future meetings, this is what I /think/ I heard - we
didn't make or pass any motions:
- For the time being, the need for working sessions approximately quarterly seems to be real. However we will not take this as cast in stone forever, but continue to evaluate that there's actually a need.
- 2. Two days is somewhat preferable to three; for those who are 100% on LSB it's not that big a difference but for those who are breaking away from other tasks to attend it's quite a big difference. Since we want to encourage wider attendance we need to account for that.
- 3. If we're meeting four times a year, at least one of those meetings should be outside the US. In future we should look at LSB meetings as a way to get more people involved and feeling they have a chance to propose ideas - and use the scheduling to promote that.
- 4. We reaffirmed that if we schedule a meeting during a popular conference, we can't do much more than a few hours of status review during the conference. A mini-conf before the conference (or after) might work, but some expressed that this made the time commitment too long - yes it saves on travel costs but that much time away from the office might be too much. Also this miniconf idea has become popular, so many are already spending a full seven days on conferences, we can't extend that reasonably.
- 5. To make the meetings more widely useful we need to do more handling of bugs on a regular basis. This meeting ended up doing a lot of the bug triage that should have been done during the last few months; while there was no way around it this time - it had to be done - in future most of that work should already have been done by the responsible subgroups before we come to a face-to-face. This is not a suggestion - it's the responsibility of each subgroup to handle this as part of their work.
As to specifics, we think we need to meet in early June to wrap most details for LSB 3, and in the first week of Sept. several LSB members will already be meeting for an ISO ballot resolution session and we can combine that with an LSB meeting. A final 2005 session would then likely be early December, but we didn't say anything specific about that.
Tantatively we're looking at the Portland, Oregon area for the June meeting, as there are several possibilities for hosting, but the Bay Area (Calif.) remains an option as well. For the Sept. meeting, the ISO Ballot resolution is already set for Singapore. If it's absolutely impossible for enough "regulars" to make it there we might have to separate those, but it's the plan of record today to combine them.
Update: the ISO meeting was set for Aug 29 - Sep 2, but it became clear a general meeting wasn't going to work out here.
[[[AustinNotes2]]/Comments User Comments]