Minutes from August 25 Release Wranglers call

[Paul Anderson]

Attendees
---------
Kevin Martin
Egbert Eich
Paul Anderson
Leon Shiman
Stuart Kreitman
Dan McNichol



Release Status
--------------
The release is proceeding, but will be delayed.  Kevin sent
out e-mail on Monday, but hasn't listed a new target date, as
there are still some things to work out.


Commit access
-------------
Kevin would like broaden the list of people allowed to checkin
at this time to reduce his workload.  There are some defects
with patches applied that other people are capable of checking
in.  It was decided during a previous call that only Kevin 
would be checking in code at this time, so he wanted to bring
it up for discussion.  Egbert noted that it would be good to
get additional eyes to review some of the changes before they
are checked in.  This would not open the door wide open, but
it would allow people with expertise in a specific area to
check-in approved changes without Kevin's involvement.  If
Kevin has to be involved with every subsequent commit for the
release, it will slow the release down, which nobody wants.

After some discussion, it was decided that we don't want to
add a lot of new process around this, and that people who are
givin commit access should use common sense.  Nobody had any
problems with allowing limited commit access to resolve
release-related issues at this time.


Bug list
--------
Reminder: Main release bug is #351.

Other open blockers are:

 #474: Problem with the Radeon driver with how video memory is allocated
       Nobody investigating at this time.  It needs investigation and a patch
        
 #999: Release notes bug
       This is a placeholder for bugs that need to be documented and for
       discussion of those working on various documentation issues.
       Several people are working on various portions of this.  We need to
       be sure that all of the documentation builds properly, etc.  There
       are no plans to change away from SGML or the current tools for
       this release; we can look into that after this release completes.

#1029: Hard failure
       The consensus is to leave the test in there.  There are some
       file systems and OSes for which this test will fail in all cases.
       For OSes that don't support this (and will see the test fail),
       the code can be compiled out.  Individual OSes can change their
       configuration files appropriately, although this does not solve
       the problem for those file systems that do not support this.
       Kevin will add a comment to bugzilla to see if Torrey Lyons can
       submit a patch that will make the test friendlier in some cases.

#1053: Artifacts when using composite
       This only happens when composite is enabled.  People would like
       to get this fixed in the release, even though Composite will not
       be enabled by default.

#1083: Problem with use of lndir 
       Changes stat() to lstat() so that you can't create directory
       symlink that is infinitely deep.

#1099: Placeholder for all bugs held open for the release, but not blocking 
       Only two bugs in this list now.  If they get fixed, we'll include
       them.  If not, they won't block the release.

#1101: Background pixmaps use wrong origin in child widgets of redir. windows
       Only occurs when composite is enabled.  Needs investigation. 

#1119: Build breaks with UseInstalled and ProjectRoot set
       This has been around for quite some time.  This was caught because
       the success criteria included build tests.  Egbert will take a look
       at this and review the patch that was supplied.

#1138: Another composite issue
       A patch is available, but may break something else.  Eric will
       be looking into this further.


At this time, bugzilla went down, so we couldn't discuss defects any
further.


We'll cover the remaining agenda items on Friday's call.


Next call:
----------
Friday, August 27.  Same time.


Best regards,

-paul