Core Minutes 8/24/2004
Present: Joanne Bogart, Toby Burnett,
Jim Chiang, Seth Digel, Richard Dubois, Warren Focke,
Riccardo Giannitrapani, Berrie Giebels, Navid Golpayegani, Traudl Hansl-Kozanecka, Heather
Kelly, Matt Langston, Michael Kuss, Pat Nolan, Chuck
Patterson, James Peachey, Dirk Petry, Leon Rochester, Tom Stephens, Tracy Usher
-
Announcements: (Richard)
He will be attending the DPF meeting in
Riverside Thursday-Tuesday, giving a talk on GLAST status on Tuesday.
- Science Tools Core: (Jim)
His write-up
speaks to several issues of naming and organization. There was plenty to
discuss [most of which I missed. ed.]
- Jim asked that we find a way to fix up the setting of the user and
system paths in the CMT requirements files for the parameter files. He
demonstrated how the resulting environment variables do not get set
absolutely correctly. Traudl remarked that paths starting /a/sulky/..
(rather than /nfs/farm...) were probably due to Jim's use of v1r14.
This has been fixed in later versions of CMT.
- Toby asked Traudl why the InstallArea appears in the path for pfiles.
Traudl responded that InstallArea will always appear by default but we can
switch it off per entry in CMTPATH.
- James Peachey went on to explain the purpose of the user and system
paths for the parameter files. The user path is there to retain the values
used for the last run. One should also note that if the modification time
of the system version is more recent, then the contents of the system
version of the parameters will clobber anything defined in the user area.
This can happen when one installs a new version of the code and your user
area points to older files.
- In response to Jim's request that PIL be distributed with readline,
Heather asked if we knew how to get a PIL build with readline. James
Peachey responded that by default PIL builds with readline on - he was
surprised that it was not there now. This will be fixed when James supplies
a new version of PIL in the next few weeks.
- Then a little discussion of ScienceTool naming convention.. where the
original suggestion was st_*.. sounds like we will stick with that. James
prefers to stay away from mixed case in many circumstances, but especially
for executables.
- There was some discussion of how to regulate what goes in the proposed
st_facilities package. Richard suggests package owner; Toby pointed out GR
facilities doesn't really have an owner; perhaps for st_facilities it should
be the architect.
- James pointed out that, in selecting some names, we need to consider
which packages, such as EventBin, are sufficiently generic to be used for
data other than GLAST data, and choose names with that in mind.
- Richard remarked that, if names are to be changed within ScienceTools,
the time to do it is NOW.
- Install area: (Traudl) See
her
presentation. Earlier documents were updated to reflect the present
status:
Report Feb 2004 and
Migration Plan. Improvements
include the simplification of the install of public header files and the
addition of the install of scripts.
An example showing all cases of data files is available on CVS, pkg examples/exInst.
[The first real life example (AcdRecon) was committed to CVS post-meeting.]
Recap of upgrade status: The latest CMT v1r16 should now become the
default; it has been tested for all releases. Gaudi v14r0 or later is needed
when the jobOptions are installed in several InstallAreas; Gaudi v15r5
requires the update of the converters (see JIRA
Gaudi-1).
Points mentioned in discussion include:
- (Traudl) whether include files are in the install area is up to the
individual package
- (Traudl) newer Gaudi supports a job options search path
($JOBOPTSEARCHPATH)
- (Toby) He would like to remove AIDA and HTL from the list of independent
externals and incorporate them with Gaudi, since only Gaudi uses them. Any
objections? Traudl mentioned that AIDA is only an interface which is useful to
connect to different histogramming packages, and that it is implemented in the
latest Gaudi with -no_auto_import.
-
Symbolic links for Windows: (Matt) They
are available, and have been for a while, as he describes here, along with
links to resources making this feature easier to use within Explorer.
See especially this
shell-shocked article which does an excellent job of describing how to use
the feature and what the pitfalls are.
-
Release Manager: (Alex,
Richard, email) The VRVS gods wouldn't let Alex attend the meeting; we'll get
the full story next week. Schedule should go something like this:
- Aug. 30: first edition of new backend perl modules
- Sept. 8: migrate current functionality to the new backend; start on
windows
- Setp. 20: new frontend; will permit more administrative control of builds.
May be ready to demonstrate Windows builds.
Plan to have a code review soon after Heather returns in mid-September.
Flight integration news for the week:
AOB
- Traudl suggests you might want to take a look at the
ICHEP04 program.
- Thank you, Heather, for filling in for
a tardy colleague.
J. Bogart Last Modified:
01-Jun-2010 15:45:42 -0700