Core Minutes 2/8/2005
Present: Joanne Bogart, Anders Borgland, Toby Burnett,
David Chamont, James Chiang, Johann Cohen-Tanugi, Seth Digel, Richard Dubois,
Warren Focke, Riccardo Giannitrapani, Tony Johnson,
Heather Kelly, Michael Kuss, Matt Langston, Pat Nolan, Chuck Patterson,
Dirk Petry, Leon Rochester, Tom Stephens, Tracy Usher
- Announcements: (Richard)
The various problems and ensuing delays in assembling the
detector may mean that we will be forced to descope. Configuration
being considered would be missing trackers at the corners. (Toby) To
first order, this just reduces effective area
proportionately.
[On a much happier note, see also this
announcement. ed.]
- Data Handling: (Tony)
- System Tests There was a recent new minor release including
several bug fixes. Expect another such soon. Also right around the
corner is the move to the glast-ground server.
- Pipeline Highest priority is to add the ability for one task to
get its input data from a different task. Matt and Dan will be working
on this together. Matt will extend the XML description to support this.
(Richard) Timescale? (Tony) Unknown at this point, though it doesn't seem
like it should take very long
- JIRA, Confluence upgrade We have been running newer versions of
both utilities in a test mode with some users. Plan is to upgrade tomorrow
starting around 4 PM; outage could last about 3 hours.
Tony will contact Ric Claus, a heavy user, to make sure he has no objections.
We will be advancing to significantly newer versions of both products; there
might be glitches.
- SWIG and so forth: (Jim)
In order to simplify the process of making Science Tools functionality
available to Python (just as rootcint is used to make our shared
libraries available to interactive ROOT) Jim has
made a new package, SwigPolicy.
From its mainpage:
SWIG,
the Simplified Wrapper and Interface Generator, provides an easy way
of exposing C and C++ code to higher level languages, such as
Python.
This package provides the infrastructure for exposing SAS C++ code as
modules that can be readily imported and used in the Python environment.
The whole thing
is worth reading.
(Toby) Use of SwigPolicy
adds two custom build steps (one to translate
CMT macros, another to actually run SWIG) and extra outputs which cause
a certain amount of clutter. He suggests creating separate very
small container packages associated with each package to be SWIGified or (Jim's
idea) just a single container to do them all. (Riccardo) Are we planning to
also wrap Gaudi packages? BOOST is also intended for this purpose. Riccardo
has had some experience with both (BOOST and SWIG) and finds SWIG easier to
use. (Jim) It would be possible to mix and match. (Toby) Use of SWIG does
have some implications for the design of packages. For example, it doesn't
support nested classes.
- GlastRelease v6r0: (Toby) Latest Tkr code moved to
HEAD; looks ok. (Richard) Zach is finishing up rest of CAL work; close
to done.
- RM: (Navid) Last of major requests
for Linux RM done:
- email messages to package owners upon failure
(but several packages have no owner)
- Put failures at top of list.
(Toby) has one more request: that there be a clear indication when checkout of a
package fails, so that subsequent failures of packages using that package are
easier to understand. Navid's response:
When the checkout of a package fails, cmt does not tell you any error at all as you can see from checkout.txt.
If I just marked it as failure, the user would see a checkout failure but the log would indicate no problem at all.
To make sure the user isn't confused what I've done is: 1)
The RM marks the checkout as a failure and 2) I append the output of check.txt to checkout.txt.
What this also means is that checkout step of RM will mark ANY cmt configuration issue as a checkout failure. So a syntax error in your requirements file means the RM considers it as a checkout failure.
Can now connect to GLAST06 (though only from home because of Goddard's
prejudice against VPN), the machine which will be doing Windows builds.
(Richard) The Web interface is slow. The server, verified to be
SLAC's production server, is powerful enough that it is most likely
not responsible. When Navid makes the same sql queries
used to produce the web pages on the machine hosting the MySQL dbs, response is
plenty fast, so the lag is from somewhere inbetween.
∫ Flight(w, ..) dw
-
EM: (Heather)
The latest tag, v3r0407p8, passed System Tests, but there is a problem
taking FITS files to digis (ldf format to digis is ok) which she will
be investigating. There is a new version of the LDF library, including
mods to the GEM data which I & T wants, so she will be incorporating that
into a new tag soon. I & T is not in a rush for new Tracker code.
Heather is aiming for a tag next week incorporating it. And, when
the new Cal code is ready, she would like to incorporate that
all by itself, if possible, without other changes.
-
ToT(Joanne) The minimal set of necessary
pieces are there, though one step (translation from XML to ROOT) has
to be done by invoking the translator program by hand. (Leon) expects
to get back to this approximately now.
- Callable interface to metadata dbs: (Joanne)
Refactoring of calibUtil is in progress. She hopes to get some
guidance from clients, in particular David Smith, on what form the interface
should take to be most useful.
Coming Soon
(Riccardo) New little collaborator to join Údine group
around August, 2005.
J. Bogart Last Modified:
01-Jun-2010 15:47:35 -0700