LSB conf call notes for 2008-11-19
Submitted by LSB List on Wed, 11/19/2008 - 17:30.
Posted to the lsb-discuss mailing list by: Jeff Licquia
Kraft, Brian Proffitt, Kay Tate, Russ Herrold, Jiri Dluhos, Alexey Khoroshilov, Ron Hale-Evans, Ted Tso. LSB 4.0 beta 2. Jeff: Decided on doing a beta 2, had autobuilder issues Monday and Tuesday which are now cleared up. Will be doing the push of SDK, tests, appbat today. Spec pushed to betaspecs. The SI will be available for more architectures this weekend; should have all but ia64 by Monday. Issues. Mats: azov test suites were segfaulting. May have been related to Xvfb, but sometimes works, too. Jeff: bugs? Mats: yes. P1 bugs. Jeff: #!, vtable, stack checker bugs still open. Mats: review is the only thing left. Jeff: should call a drop-dead date? Mats: not convinced that absence of comments means approval. Jeff: yes, but people can always file new bugs if they don't like the wording. Jeff: bug 2386. Mats: although lsb-release can return multiple elements, parsing that is not ideal. If we wanted to return just 4.0 items, we could add spec wording to the effect that 3.x is assumed. No preference. Ted: do apps depend on this? Mats: would be surprised if there was. Ted: deprecation might be an issue later. Jeff: not until LSB 6. Don't have a preference either. Ted: flip side, if we are explicit now, we'll discover the bugs sooner rather than later. Kay: other competing requirement: might choose to turn on optional features if newer LSB is available. Ted: for that scenario, might also export what modules are available. Can also help trial use. Mats: format was designed to be able to describe modules, too, but don't export more than core and desktop. Could add trial use support, too. Kay: smaller ISVs would probably rather get back an answer that's a general statement instead of looking at the list. Jeff: do both? Ted: only issue is to get that feature into the distros; some use our sample, some don't. If we do modules, people are forced to pick from the list no matter what; if we care about ISVs wanting just a summary, we'll have to add it as a separate option. Jeff: package deps. lsb-core-ia32, for example. Have to drop versions for Debian; maybe drop entirely, or drop versions entirely? Russ: won't that break yum and apt? Jeff: currenly, only done for Debian. Ted: does anyone use them? Mats: no, not apart from us, which is irrelevant. It's part of the ISO spec. Sort of in a legacy mode. Jeff: probably should just do the Debian workaround, then. Jeff: chicken-and-egg problem with lsb package and lsb-release; will everyone issue updates? Mats: probably unlikely. Ted: before we release, no one knows if they'll be 4.0-compliant. Will have to update something, most likely. Mats: we don't make our stuff depend on the latest LSB unless necessary. Jeff: Mats: timeline? Jeff: could do a rc as early as December 3. Too optimistic? Mats: don't think so, but has been wrong before. Ted: let's try for the 3rd, and slip if we have to. Ron: there are spec additions not in beta but need review. Reentrant random functions. See http://dev.linux-foundation.org/snapshots/booksets/LSB-Core-generic/LSB-... for the snapshot versions.
Kraft, Brian Proffitt, Kay Tate, Russ Herrold, Jiri Dluhos, Alexey Khoroshilov, Ron Hale-Evans, Ted Tso. LSB 4.0 beta 2. Jeff: Decided on doing a beta 2, had autobuilder issues Monday and Tuesday which are now cleared up. Will be doing the push of SDK, tests, appbat today. Spec pushed to betaspecs. The SI will be available for more architectures this weekend; should have all but ia64 by Monday. Issues. Mats: azov test suites were segfaulting. May have been related to Xvfb, but sometimes works, too. Jeff: bugs? Mats: yes. P1 bugs. Jeff: #!, vtable, stack checker bugs still open. Mats: review is the only thing left. Jeff: should call a drop-dead date? Mats: not convinced that absence of comments means approval. Jeff: yes, but people can always file new bugs if they don't like the wording. Jeff: bug 2386. Mats: although lsb-release can return multiple elements, parsing that is not ideal. If we wanted to return just 4.0 items, we could add spec wording to the effect that 3.x is assumed. No preference. Ted: do apps depend on this? Mats: would be surprised if there was. Ted: deprecation might be an issue later. Jeff: not until LSB 6. Don't have a preference either. Ted: flip side, if we are explicit now, we'll discover the bugs sooner rather than later. Kay: other competing requirement: might choose to turn on optional features if newer LSB is available. Ted: for that scenario, might also export what modules are available. Can also help trial use. Mats: format was designed to be able to describe modules, too, but don't export more than core and desktop. Could add trial use support, too. Kay: smaller ISVs would probably rather get back an answer that's a general statement instead of looking at the list. Jeff: do both? Ted: only issue is to get that feature into the distros; some use our sample, some don't. If we do modules, people are forced to pick from the list no matter what; if we care about ISVs wanting just a summary, we'll have to add it as a separate option. Jeff: package deps. lsb-core-ia32, for example. Have to drop versions for Debian; maybe drop entirely, or drop versions entirely? Russ: won't that break yum and apt? Jeff: currenly, only done for Debian. Ted: does anyone use them? Mats: no, not apart from us, which is irrelevant. It's part of the ISO spec. Sort of in a legacy mode. Jeff: probably should just do the Debian workaround, then. Jeff: chicken-and-egg problem with lsb package and lsb-release; will everyone issue updates? Mats: probably unlikely. Ted: before we release, no one knows if they'll be 4.0-compliant. Will have to update something, most likely. Mats: we don't make our stuff depend on the latest LSB unless necessary. Jeff: Mats: timeline? Jeff: could do a rc as early as December 3. Too optimistic? Mats: don't think so, but has been wrong before. Ted: let's try for the 3rd, and slip if we have to. Ron: there are spec additions not in beta but need review. Reentrant random functions. See http://dev.linux-foundation.org/snapshots/booksets/LSB-Core-generic/LSB-... for the snapshot versions.