Core Minutes 11/09/2004
Present: Joanne Bogart, Toby Burnett, David Chamont,
James Chiang, Johann Cohen-Tanugi, Seth Digel, Richard Dubois, Zach Fewtrell,
Warren Focke, Riccardo Giannitrapani, Berrie Giebels, Navid
Golpayegani, Heather Kelly, Michael Kuss, Chuck Patterson, Leon
Rochester, Tracy Usher
-
Tagging in MRvcmt: (Riccardo) He and Marco
have implemented comprehensive support. To use it, select a package and click
the tag icon or
find it in the CVS menu. You will be presented with a dialog which makes it
easy to increment version, revision, or patch, or to type in an arbitrary tag.
For container packages only the cvs tag command is used. For non-container
packages cvs rtag (which makes a tag which is a snapshot of the package
HEAD, regardless of what might be in your working copy) is the default, but
tag is also available. In all cases MRvcmt will update the version line in
your requirements file and will optionally allow you to edit release.notes
[why is it optional? ed.]
There was some additional MRvcmt discussion.
(Richard) How about a new feature, allowing one to right-click on a file in
the package browser and have an option to commit just that file to CVS? (Toby)
On Windows TortoiseCVS provides a
very nice interface to CVS, including ability to select files to commit.
It also makes it easy to issue cvs -n update, a very useful command.
(Joanne) You can do this easily from MRvcmt also; it's one of the items in the
CVS menu. (Richard) ..wanted to know if anyone had a solution to problem with
spawned terminal windows with no usable delete or backspace key while running
MRvcmt on Linux from a Windows desktop. (Joanne) ..later rediscovered old solution
to similar problem: include the
following in your .Xdefaults
XTerm*VT100.Translations:#override \
None <Key>BackSpace: string(0x08)
-
RM requests: (Toby)
- We're missing output for GlastRelease v4r5p1. (Joanne) This was probably a
casualty of an SCS problem with nfs disks. It should be possible to
regenerate the output.
- We had agreed that RM should
automatically create branch tags for each release, but we still don't have
it. (Joanne for mute Navid) Although this might involve only a line or two of
code, ramifications need to be thought through carefully. It involves a write
to the CVS repository, not something RM has done up to this point. [See also
infrasoft thread.]
-
Glast0N outage: (Richard)
We are still without Glast01-Glast04, which means we're without Release
Manager, System Tests, MySQL calibration database, Tomcat server, Pipeline,...
Richard has made a request to SCS that they plug these boxes in somewhere
short-term if need be since we are severely handicapped. (Seth) How about
moving the functionality to other machines? (Richard) Some data (e.g.,
MySQL databases) are local to a particular machine. Then there is
configuration: portable in theory, but not always so easy to move in practice.
To some extent the Glast0N machines can sub for each other, but in this case
we lost all of them. We need to make arrangements of some kind so we won't be
vulnerable to this sort of outage in future.
- Pipeline:
(Richard for Dan) Dan has a work-around for pipeline problem of last week
(batch system reports "done" when it isn't quite) which looked good after some
testing, but with the Glast server outage he hasn't been able to put it in.
(Richard
for Richard) Has been working on all-gamma pipeline. He is trying to arrange
it so that all the configuration can be done in a single job options file,
without use of environment variables. In support of this, he made a small
change to GlastRandomSvc to provide string version of run number.
- Optimized builds: (Julie) Don't we
want an optimized build of Gleam before embarking on a serious all-gamma run?
(Heather) ..has been working on getting an optimized build of EM. First step, to obtain optimized builds of external libraries,
is now complete. (In a
couple non-crucial cases -- zlib, OmniOrb -- she just made a soft link to
existing libraries, which may or may not be optimized.) In email to Heather, Alex
stated he expects to be
able to reinstate optimized builds for EM; GlastRelease shouldn't be any
different [but until we're past the Glast server outage nothing can be done].
- ROOT upgrade: (Heather)
Work is in
progress on moving us to ROOT 4.00.08, the current production version. Testing
so far (Linux only) has gone smoothly. This version has a flaw (nothing new;
older versions also have it): it is generally possible to create a shared
library on the fly from interactive ROOT while reading a ROOT file, without
having to build the library in advance, however this doesn't work for classes
containing std::vector. In particular, it doesn't work for current digi
and recon files. This problem is fixed in the development version of ROOT, but
Heather is inclined to just move to 4.00.08 this round, waiting until
the current development version morphs into production before upgrading
further.
- Xerces upgrade: (Joanne) Have binaries
for both platforms and have modified versions of almost all affected packages.
Might be ready for committing and tagging by tomorrow.
- ROOT conversion service: (David) He
thinks it will be ready by next week.
∫ Flight(w, ..) dw
- EM status:
(Heather) We have a tag which is a candidate for I&T use. Waiting for
System Tests results. If these look OK, we will be in a position to bring in
new external libraries [see ROOT and Xerces items above] and to make other
significant upgrades.
Coming Soon
- J2EE Review: Date will be Nov. 23;
time TBD. This review is different from the usual code review in that the
reviewers (all external to GLAST) are charged to evaluate feasibility and
appropriateness of the technology. See more details on the
review home page.
- JIRA: Will look at open issues
next week; it's been a while since the last time.
J. Bogart Last Modified:
01-Jun-2010 15:47:58 -0700