Difference between revisions of "GlueX Offline Meeting, March 23, 2010"

From GlueXWiki
Jump to: navigation, search
m (Reverted edits by Marki (Marki); changed back to last version by Wolin)
Line 61: Line 61:
 
==[[Versioning of Calibration Constants]]==
 
==[[Versioning of Calibration Constants]]==
  
Mark reviewed the interim scheme (link above) that has been implemented. Elliott made a suggestion involving a service-oriented architecture. Details were not clear to the secretary.
+
Mark reviewed the interim scheme (link above) that has been implemented. Elliott made a suggestion involving a [[CLARA and Service Oriented Architectures|service-oriented architecture]]. Details were not clear to the secretary.
  
 
==[[media:FCALClustering.pdf|FCAL clustering]]==
 
==[[media:FCALClustering.pdf|FCAL clustering]]==
Line 86: Line 86:
 
* Matt thought that Notes should include the definitions of external, tunable parameters as well as describing algorithms.
 
* Matt thought that Notes should include the definitions of external, tunable parameters as well as describing algorithms.
 
* Mark will likely make a proposal to the group at a future meeting.
 
* Mark will likely make a proposal to the group at a future meeting.
 
==CLARA and Service Oriented Architectures==
 
 
CLARA is the CLAS12 framework for implementing a Service Oriented Architecture, or SOA.  SOA's are typically client/server based software systems that employ a number of servers distributed on the network, where more than one server can provide the same service.  Servers are listed in a central registry so clients can choose which server to use based on various criteria, including how busy they are.  In order to perform their work servers may also act as clients and request services of other servers.  Execution of the algorithm the system implements is thus distributed among many servers, similar to how an algorithm is distributed among many functions or methods in a traditional program.
 
 
In any SOA-based system it is important to balance the granularity of the service, i.e. how much work each server performs, with the overhead needed to supply the server with whatever information it needs and how long it takes to receive the result.  For example, it probably does not make sense to implement a "square root" service.
 
 
Currently CLAS12 and Hall D's thinking differs on the optimal level of service granularity.  CLAS12 tends towards breaking up their reconstruction algorithm among many servers (tracking service, cluster finder service, etc).  Hall D leans towards much coarser services (supply magnetic field, supply calibration constants, perform full event reconstruction, etc).  CLAS12 and Hall D are currently implementing test services to learn more about this.  The Hall D service will use CLARA but internally will use the JANA framework.
 
 
  
 
==New Action Items==
 
==New Action Items==

Revision as of 09:25, 25 March 2010

GlueX Offline Meeting
Tuesday, March 23, 2010
2:00 pm EDT
JLab: CEBAF Center F326/7

Telephone

To connect by telephone:

  1. dial:
    • 800-377-8846 : US
    • 888-276-7715 : Canada
    • 302-709-8424 : International
  2. enter participant code: 77438230# (remember the "#")

Video Conferencing

A) ESNet: 8542553
B) EVO: direct meeting link

Slides

Talks can be deposited in the directory /group/halld/www/halldweb1/html/talks/2010-03 on the JLab CUE (you have to be a member of the "halld" Unix group). This directory is accessible from the web at https://halldweb1.jlab.org/talks/2010-03/ .

Reference Material

Material on Splitoffs

Agenda

  1. Announcements
    1. New src release, release-2010-03-10: Mark
    2. New JANA release, version 0.6.1: David
    3. src -> sim-recon transition made: Mark, Beni
    4. BMS change: binaries depend on libraries: Mark
    5. Partial wave analysis, initial contacts: Craig
  2. Review of minutes from the last meeting: all
  3. Versioning of Calibration Constants: Mark
  4. Clustering in the Calorimeters: FCAL clustering Mihajlo
  5. Material on Splitoffs: Curtis
  6. Filtering events from bggen: all
  7. Pointers to cathode clusters in FDC pseudopoints: Mark
  8. Documentation Structure: Mark
  9. Action Item Review: all

Minutes

IU: Mihajlo Kornicer, Ryan Mitchell, Matt Shepherd, Claire Tarbert
JLab: Craig Bookwalter, Mark Ito (chair), David Lawrence, Yi Qiang, Sascha Somov, Simon Taylor, Elliott Wolin, Beni Zihlmann

Announcements

  • New src release, release-2010-03-10: Mark asked if anyone has had problems. None were reported.
  • New JANA release, version 0.6.1: David told that some changes in this release were prompted by Elliott's work on the DANA-to-EVIO conversion packages. JANADOT (gives times spent in factories) has improved significicantly as well. Mark suggested some documentation on how to generate the JANADOT report. Elliott suggested a new release so he can access pre-built versions using this version.
  • src -> sim-recon transition made: Mark noted that the change has been made. No problems were reported by the assembled. Beni has been working on updating documentation.
  • BMS change: binaries depend on libraries: Mark mentioned the recent change made to the binary making makefile (Makefile.bin). Now binaries should automatically be remade when libraries, upon which they depend, have changed. The should largely remove the need for invoking make with the "relink" target.
  • Partial wave analysis, initial contacts: Craig has been in contact with Matt and Curtis on this. Discussions are on-going. Matt and Craig agreed to get together to get Craig going with the IU PWA code.

Review of minutes from the last meeting

We went over them. Elliott volunteered to check-out and run Dave Heddle's GlueX event display. He also mentioned that a summer student may be working on this.

Versioning of Calibration Constants

Mark reviewed the interim scheme (link above) that has been implemented. Elliott made a suggestion involving a service-oriented architecture. Details were not clear to the secretary.

FCAL clustering

See Mihajlo's slides (link above) for details. There is NIM paper written on this subject. We agreed that the paper should be linked from the FCAL detector page.

Matt pointed out that the approach Mihajlo described is oriented toward electromagnetic showers. It might not work well for hadrons that interact.

David told us that currently the extra photons are coming mainly from the BCAL. Matt reminded us that BCAL reconstruction was taken over from KLOE code. We do not have a GlueX "owner" for this code. A GlueX Note does exist on this subject.

Material on Splitoffs

We peeked at the page of collected references (link above) from Curtis and Matt. David has attempted the Dolby thing. This is work in progress.

Pointers to cathode clusters in FDC pseudopoints

Mark discovered recently that these are not in the DFDCPsuedo class. Simon volunteered to put them in.

Documentation Structure

Mark outlined some ideas about documenting our software (see link above). The overall question is whether we should have a formal policy about documentation and if so, what should be in it. We had a discussion:

  • Beni pointed out that documentation can also be put into code as comments, and can be extensive.
  • Matt thought that Notes should include the definitions of external, tunable parameters as well as describing algorithms.
  • Mark will likely make a proposal to the group at a future meeting.

New Action Items

  1. Document on JANADOT -> David
  2. "RedHat 5.3" link to "CentOS" for JLab builds -> Mark
  3. Check-out and run ded. -> Elliott
  4. Provide description of SOA for calibration constant versioning for the minutes. -> Elliott
  5. Link the FCAL algorithm NIM article from the FCAL page. -> Mihajlo
  6. Find an "owner" for the BCAL reconstruction code -> Mark, Matt
  7. Schedule a discussion of filtering events from bggen.
  8. Put pointers to cathode clusters in the DFDCPseudo class. -> Simon
  9. Re-raise the issue of a documentation policy at a future meeting. -> Mark