Kyle brought up the terms skimming, slimming, and thinning during
a PAT talk at software week. I've been trying to understand the implications
on our tool set. Here's my take.
, n., the extraction of events of interest from the data store.
- this is more or less what the TAG's are made for, and what we've
prototyped with our extraction utilities/browser.
, n., limiting the itemlist during a Skim to a specified list of
container#key combinations useful for the physics.
- this could
be done if we had an actual list of what container#key
combinations were in AOD,ESD,... It's not as easy as it should be. The
list cannot be compiled from the output because the job options may
have container#*. To get it from the job options requires actual execution
of python, since it is built from multiple fragments, and varies from release
to release. It would be nice if it were saved in the metadata system for
, n., limiting the objects in each container in the itemlist based
on physics principles.
- I'm not sure we can help here. Slimming just selects containers, but
thinning actually alters their state and does require some code to build
the list, which may be hard to generalize. I think we should give good
pointers on how to do it, but leave everything else to the physicist.
Summary: We should provide the best skimming help we can, and help
with slimming if we can.
New Oracle DB format and Connecting to CERN Oracle
Script to dump attributes from a Collection:
This is the script I use to dump the item list. If you have set up an
Atlas release then do
- You need to have a RecOpts
.py in your directory.
- The output is more verbose that I would like, but I haven't found the flags to reduce it.
- I don't know that this will work with all releases.
I've just updated pathena on CERN CVS. If you have
= [ "streamtest_exclEle1373032551.root" ]
in your jobO and the root file exists on the current directoy,
pathena --shipInput ...
will export the root file to WNs and Athena will run on the file.
You can use --nEventsPerJob=??? when you want to split the job.
� � Tadashi
From: Jack Cranshaw <email@example.com>
Subject: TAG toys
Date: Thu, 14 Jun 2007 10:33:49 -0500
This is a prototype page with limited functionality for doing a query
on a relational TAG database for the streamtest. It allows a certain
amount of browsing. It also allows the user to generate a root TAG file
based on the selection. For example, the one here
What would be the easiest way to use this as input to a pathena job
since I want to use DA, but it's not in a dataset?
Other from Jack
Subject: loaded a file
Command I was using
~/maxidisk/workarea/data/inclEle/tag % CollListAttrib
.exe -src streamtest1 RelationalCollection
You can find the files that I've been transferring for the streams test in
I've been running a script in
to do the loading by hand.
I forgot to send the connection information for the CERN db.
Please feel free to add, but it might be nice to have this for the discussion.
On Jun 14, 2007, at 7:55 AM, Helen McGlone
'> Hi all,
- 09 Aug 2007