[lsb-discuss] unoffficial LSB conference call minutes for 13 Mar 2012

R P Herrold herrold at owlriver.com
Wed Mar 13 17:10:57 UTC 2013

LSB teleconference

Wednesday, 13 Mar 2013 @11:00 to noon, US ET

Where:	(605) 715-4920 Access Code: 512468

Gobby:	gobby -j lsb-temp.pmman.net:6522 with passwd: LSB

IRC channels are:
 	#lsb-meeting at freenode.net during meetings
 	#lsb at freenode.net most other times

in the form : name (organization)  [irc_nick]
Jeff Licquia (LF)  licquia
Mats Wichmann (Intel)  mwichmann
Russ Herrold (Owl River)  orc_emac
Glen Petrie (Epson)
Rick Troth (Velocity Software) rickt

Stew Benedict ()  stewb
Robert Schweikert (SuSE)  robjo
Denis Silakov (ROSA Labs)  denis_silakov
Alan Clark (SuSE)  AlanClark
Darren Davis (SuSE)
Jiri Dluhos (SUSE)  jdluhos
Alexey Khoroshilov (ISP RAS)  Alexey
Carlos O'Donell (Mentor Graphics) odonell

Kay Tate (SuSE)  ktate

Posted Agenda:
  - Stable update status.

  - Issues with LSB 5.0 changes, build slaves, targets for 
building, etc.

  - C++ 11?

New business:

Meeting opens at 11:00 with:

Updates did not make out out by today, but Jeff is close

There's been a need to deploy a new builder for x86_64 due to 
a new symbol (memcpy @ GLIBC_2.14) appearing in LSB in the 
course of "uplift libc / bug 3470, which caused fails on the 
old builder, which had only glibc 2.9.  It's had some hiccups, 
as one might expect; including two issues that have been 
"surprising" - rpmbuild is not munging tarballs so patches can 
apply even if files to be patched are read-only; and in a 
couple of places where we downloaded tarballs from 
SourceForge, the url was encoded with mirror information, new 
to this instance (wget build option? something else?), that 
encoding appeared in the filename so to the build it appeared 
never to have been downloaded.

Why did we run into this problem? We do try to build for 
conservative targets, e.g. SDK itself is built to require old 
version.  But tests have to target the version they're 
testing, and in some (many?) cases the test suite build builds 
code which is then executed in continuing the build.  memcpy 
was an unfortunate symbol to have this, because a lot of code 
uses it, largely "we've been unlucky". We seemed to run into 
it in the test /framework/ builds, even.

IA64 again --- Both SuSE and RHEL have dropped it going 
forward; no RPM based distribution has it at the latest 
levels.  Should time be spent re-working the builders to work 
under Debian? (which was jokingly noted as likely to support 
IA64 until the end of time).  This however loses the RPM 
'BuildRequires' information.  Mats points out that these 
should be in the lsb-* namespage.  Robert reminds the call 
that in our prior discussion, if IA64 is important to the LF, 
the proponents need to step up and participate.  This has not 
happened to date.  The takeaway was a 'wait and see' when 
stuff breaks, and do not rework builders now.

Mats asks for "decision" on some uplift bugs, what would be 
the precise set of symbols to add.

Call ends at local Noon

Weekly Bug triage:
The bug tracker is at: https://bugs.linuxbase.org/
irc-only #lsb-meeting at freenode.net
Thurs at 10:00 Eastern
Bugzilla stats as of 13 Mar: 541 open (535 last week), 536
 	assigned, 5 new
New last 90 days: 61, last 30: 28; closed last 90: 44, last
 	30: 12
Changed this week: 146 (lots of reassignments), last week: 11
Bugs that are assigned but have no milestone: 226, Rollup 
bugs: 23


Next meeting:
20 Mar 2013 11:00 to noon, US ET

gobby unofficial minutes are open for edit for at least 15 min 
after each call, and then issued into the mailing list

This document: LSB_minutes_20130313.txt

More information about the lsb-discuss mailing list