[lsb-discuss] LSB conf call notes for 2008-02-27
jeff at licquia.org
Wed Feb 27 09:38:40 PST 2008
Attendees: Jeff, Mats, Sam, George, Robert, Darren Davis, Dan Kohn,
Jesper, Stew, Marc Miller, Ted, Carlos, Russ, Kay, Markus.
Google Summer of Code announced; do we want to do this? Ted: no
guarantee of funding or completion. What projects are there that aren't
critical? George: dynamic application checking. Dan: agrees with Ted;
management resources sucked in that we don't have. Jeff: maybe next
year, after successful LSB 4 release. Dan: if we have stuff to do, we
have money of our own. Ted: also, project doesn't start until June,
complete until August; poor fit with our current schedule. Talk about
various projects; Berlin API deserves more effort. Enhancements to
alien might be a good candidate. Jeff will research. Russ has been
doing alien work; may mentor the SoC candidate.
Reimports. Mats: tools in pretty good shape. Qt 3 pretty much done, Qt
4 in progress, libstdc++ in research phase. Ted: how much time does it
take to do a reimport? Mats: run the tool, which suggests stuff to be
added, which needs to be researched. The analysis is what takes time.
ISPRAS has been doing most of the work there; perhaps several days to
look at the material. Ted: how many libs? Jeff: Qt 3, Qt 4 (several),
libstdc++, X11 (several), libgl. Mats: we can't use our data to
generate headers without the reimports. Ted: this is also the same
process to uplift? Mats: yes. Ted: if there's anything more to
automate, we can do that. Mats: ISPRAS doing the heavy lifting,
probably know about tool quality. Jeff: should we ask ISPRAS about tool
quality? Mats will keep the conversation going; Ted and Markus will
also ask for it in the status report.
glibc. Mats: discussion pretty much resolved to target glibc 2.4. Need
to identify symbols to add, both from 2.4 and from older versions.
Jeff: multiple symbol version support? Mats: yes, old policy was having
one symbol and one symbol version; this will change in 4.0. Related to
the requirement to link libpthread. Database supports it now, but there
are lots of assumptions in other code that there's only one. Ted: who
has the skills to do this, besides Mats? Mats: ISPRAS can; have been
going back and forth with them on this. May need to refactor
deprecation to tie to the symbol version instead of the symbol itself.
Jeff: strategy? pre-worry, or just do it and test to find problems?
Mats: we know which scripts assume this, so we can do some preparation.
Ted: when can we start adding multiple symbols to test this? Mats:
can create a test case easily; pthread, for example.
D-Bus. Mats: versioning issue; SLES 10 has D-Bus older than 1.0 (0.6).
Stew: RHEL looks like 0.7. Mats: nervous about pre-1.0 API in
standard. Ted: any requests? Jeff: ATSPI, HAL, other integration.
Ted: maybe push off past 4.0? Kay: BTW, have been sending app info to
ISPRAS. Ted: anyone using D-Bus? Kay: larger ISVs to start using it;
they work with the smaller ones. May be better to talk to larger ISVs
directly. Ted: meeting with ISVs next week; can bring up that question.
Kay: people should submit data to the Navigator. Darren: can be a
hard sell; nervousness about doing analysis on code. Ted: maybe we
would be more successful if blinding were easier. Dan: SAP is available
as a binary for free download. Darren: maybe if some of the data was
restricted. Ted: maybe do D-Bus 1.0 as Google SoC as a trial-use
module. Jeff: apps in Navigator seem to be mostly open-source; Skype
probably the most prominent. Ted: interest as GSoC? Sponsor? a11y?
Robert: maybe D-Bus should be medium instead of high. Ted: maybe even
low, as trial-use. Dan: intersection between 0.6 and 1.0?
ALSA. Mats: want feedback on trial-use standard. If we don't make
changes, could flip a switch. Jeff: tests? Mats: should look at that.
Maybe, ISPRAS was looking at that. Markus: ISPRAS already looking at
this, were going to do for the May code drop. Ted: have we seen any
upcoming ABI changes in bleeding-edge ALSA? Mats: SuSE ALSA guy figured
this out with ISPRAS, don't know the exact resolution. Markus will find
out. Mats: when the tests appear, we can move it over. Jeff: should
use the wiki to fill in status info.
Mats: adding libGLU is almost done; just need a published spec. Ted: is
there a spec somewhere? Mats: contained in the general OpenGL spec,
just need to make sure it's as complete as we need.
Ted: please update status page!
More information about the lsb-discuss