July 1, 2009 Software

From GlueXWiki
Jump to: navigation, search

Location

The meeting will be at 2:00 pm in CEBAF Center F326/327.

Telephone

To connect by telephone:

  1. dial:
    • 800-377-8846 : US
    • 888-276-7715 : Canada
    • 302-709-8424 : International
  2. enter participant code: 39527048# (remember the "#")

Video Conferencing

A) ESNet: 8542553
B) EVO: meeting link

Agenda

  1. Review minutes from June 17, 2009 Software meeting
  2. Announcements
    • Anyone have any?
  3. Work disk: Mark
  4. Software tag policy--when to modify the tag: Mark
  5. Removing 2-GB-file-size limit (makefile changes): David
  6. DRootGeom changes--multi-threading and ROOT (Slides): David
  7. Magnetic Field Documentation Webpage: David
  8. Proposed changes to mcsmear: David
  9. Group disk usage (work and cache for that matter): all
  10. Hall D Unix groups discussion: Elliott
  11. Tracking removed from HDGeant: David
  12. Action item review

Minutes

Present:

  • Indiana: Matt Shepherd
  • JLab: Eugene Chudakov, Mark Ito (chair), Alexander Somov, Simon Talyor, Elliott Wolin, Beni Zihlmann

Work disk

Mark shared what he has learned recently about the new work disk quota system at the Lab. It has a new "shared" space feature which can make it appear that a particular partition is changing its size dynamically.

Software tag policy--when to modify the tag

Mark led the discussion. From time to time, we are tempted to modify a tagged release when a bug fix is found, post-tagging. What to do depends on balancing the desire to have a tag have a time-independent meaning and the desire to fix something that is broken when that fix requires only a very small change. So far we have left this to the judgment of the person maintaining the tag. There was a consensus on continuing with this practice.

Removing 2-GB-file-size limit

David led us through his email message describing the change to the make system to fix the limitation on HDGeant output file size. The change was done by adding the switch in one place in one file.

DRootGeom changes--multi-threading and ROOT

David showed some slides describing the problem and options for dealing with thread safety and the DRootGeom class. There are still some kinks to work out with the table method. More on this next meeting.

Magnetic Field Documentation Webpage

David led us through the issues documented on this page. He will note on the page the appropriate use of the -Z0 switch.

Proposed changes to mcsmear

We discussed the issues raised in the recent email discussion about an intermediate stage between HDGeant and reconstruction that would add random detector effects and create appropriate HDDM output. Matt pointed out that the structure of the smeared hits is in some cases much different than the structure of the hits coming out of HDGeant. For example the BCAL hits have the energy deposited in a module as they come out of HDGeant, but not the detector response at each end. There is no place in the data model for this information at present. mcsmear remains the leading candidate for providing this function. David will take the lead in studying the changes this would require.

Hall D Unix groups

Elliott led us through his proposal for re-arranging our set of Unix groups at JLab. We agreed that he should go ahead and implement this with IT Division. Eugene noted that the newgrp command can be used to create a new shell with the default group of one's choice. Elliott also mentioned that the directory mode bits can be set to insure that all files created in that directory belong to same group.

Tracking removed from HDGeant

David led us through his email announcing this change. This means that we are now running with the straight-line approximation within a straw for calculating the drift distance in the CDC. This error is small compared with our position resolution.

Action Item Review

Sascha suggested that someone fix the not-a-number problem in the BCAL cluster code.

Mark will go through the list and mark done items as done. Next time we will discuss the task list.

New action items:

  1. Draft a policy on when to generate new hdparsim efficiencies. -> ?
  2. DRootGeom allegedly gets recompiled even if the geometry has not changed when a re-make is done. -> David
  3. Review action item list and cross out items that are done. -> Mark
  4. Further study on how to modify mcsmear -> David
  5. Document use of -Z0 options on magnetic field web page. -> David
  6. Sascha will contact Blake about the nan problem. -> Sascha

Agenda items for next time

  1. Group disk usage (work and cache for that matter): all
  2. Serializing ROOT objects and transfering them with cMsg: Elliott

Recorded by --Marki 11:02, 2 July 2009 (EDT)