<html>
  <head>

    <meta http-equiv="content-type" content="text/html; charset=UTF-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <p>Folks,</p>
    <p>Please find the minutes below and <a moz-do-not-send="true"
href="https://halldweb.jlab.org/wiki/index.php/GlueX_Software_Meeting,_March_5,_2019#Minutes">here</a>.</p>
    <p>  -- Mark\</p>
    <p>_______________________________________</p>
    <p>
    </p>
    <div id="globalWrapper">
      <div id="column-content">
        <div id="content" class="mw-body" role="main">
          <h2 id="firstHeading" class="firstHeading" lang="en"><span
              dir="auto">Minutes, GlueX Software Meeting, March 5, 2019</span></h2>
          <div id="bodyContent" class="mw-body-content">
            <div id="mw-content-text" dir="ltr" class="mw-content-ltr"
              lang="en">
              <p>Present:
              </p>
              <ul>
                <li> <b> CMU: </b> Naomi Jarvis</li>
                <li> <b> JLab: </b> Shankar Adhikari, Alexander
                  Austregesilo, Thomas Britton, Eugene Chudakov, Mark
                  Dalton, Sean Dobbs, Stuart Fegan, Mark Ito (chair),
                  David Lawrence, Justin Stevens, Simon Taylor, Beni
                  Zihlmann</li>
              </ul>
              <p>There is a <a rel="nofollow" class="external text"
                  href="https://bluejeans.com/s/21pkN/">recording of
                  this meeting</a> on the BlueJeans site. Use your JLab
                credentials to access it.
              </p>
              <h3><span class="mw-headline" id="Announcements">Announcements</span></h3>
              <ol>
                <li> <a rel="nofollow" class="external text"
href="https://mailman.jlab.org/pipermail/halld-offline/2019-March/003544.html">Moving
                    simple_email_list</a>. The scripts are moving from
                  Subversion to Git. Cron jobs run by the gluex account
                  should use /home/gluex/simple_email_list rather than
                  /group/halld/Software/scripts/simple_email_list to
                  access the lists and the script. (Reported by Mark)</li>
                <li> <a rel="nofollow" class="external text"
href="https://mailman.jlab.org/mailman/private/gluex-collaboration/2019-February/005115.html">Another
                    way to provide feedback on wiki pages</a>. You can
                  use the "discussion" tab present on all wiki pages.
                  The email will be sent to the author(s). (Reported by
                  Mark)</li>
                <li> Did you submit a Project to MCwrapper-bot that you
                  no longer need? Or does your project have enough data
                  that you don't care about the outstanding data? You
                  can now go to the dashboard and right click on your
                  project row to <b>Recall outstanding jobs</b> from a
                  project. Or <b>Declare the project completed.</b> Or
                  <b>cancel the project.</b> All these come with
                  stipulation of ownership. (Reported by Thomas)</li>
              </ol>
              <h3><span class="mw-headline"
                  id="Review_of_minutes_from_the_February_5_meeting">Review
                  of minutes from the February 5 meeting</span></h3>
              <p>We went over <a
href="https://halldweb.jlab.org/wiki/index.php/GlueX_Software_Meeting,_February_5,_2019#Minutes"
                  title="GlueX Software Meeting, February 5, 2019">the
                  minutes</a>.
              </p>
              <ul>
                <li> Thomas will be giving two talk on MCwrapper one in
                  a general session and one at the OSG user training
                  session.</li>
                <li> David reported that batch 1 of the the Spring 2018
                  reconstruction has gone through NERSC and batch 2 is
                  in progress. We are limited by the observed 2.5 Gb/s
                  rate of transfer of the raw data out of the Lab. Bryan
                  Hess is looking at why this is less than the 10 Gb/s
                  bandwidth that is advertised. With the additional 9 M
                  units we received from Hall B, we have just enough
                  allocation to finish the run. Recall that we only
                  received about 1/3 of our requested allocation.</li>
                <li> David and Chris Larrieu are looking at using an
                  allocation at the Pittsburgh Supercomputing Center. We
                  will probably try to do a batch on the JLab farm as
                  well.</li>
                <li> Thomas has submitted a test, using XROOTD to stream
                  random trigger data to OSG nodes from a UConn server.
                  We await the results.</li>
                <li> Edgar Fajardo of the OSG has advocated using
                  StashCache to distribute the random trigger data.</li>
                <li> David Lawrence will be speaking at the Machine
                  Learning Lunch tomorrow.
                  <ul>
                    <li> From the Slack channel #ml: Reminder that the
                      ML Lunch Series will meet tomorrow (Wednesday,
                      3/6) in CC F324-25 (assuming it's available) from
                      12-1PM. David Lawrence will present recent work on
                      "Machine Learning for Particle Tracking". Look
                      forward to seeing everyone!</li>
                  </ul>
                </li>
              </ul>
              <h3><span class="mw-headline"
                  id="Report_from_the_February_26_HDGeant4_Meeting">Report
                  from the February 26 HDGeant4 Meeting</span></h3>
              <p>We went over <a
href="https://halldweb.jlab.org/wiki/index.php/HDGeant4_Meeting,_February_26,_2019#Minutes"
                  title="HDGeant4 Meeting, February 26, 2019">the
                  minutes</a>.
              </p>
              <p>On Alex's ρ analysis, there were some comments:
              </p>
              <ul>
                <li> Thomas confirmed that the geometry used for HDG3
                  and HDG4 were the same.</li>
                <li> The start counter problem noted last time was fixed
                  by Richard Jones. Alex re-ran the comparisons and
                  several of the plots look to be in better agreement,
                  however...</li>
                <li> The difference in efficiency as a function of
                  energy persists. Differences remain even if the
                  comparison is done with no trigger simulation cut.</li>
                <li> Sean believes the differences are due to shower
                  energy and timing problems when using HDG4.</li>
              </ul>
              <h3><span class="mw-headline"
                  id="Report_from_the_SciComp_Meeting_on_February_28">Report
                  from the SciComp Meeting on February 28</span></h3>
              <ol>
                <li> <a rel="nofollow" class="external text"
                    href="https://indico.cern.ch/event/759388/timetable/#20190318">HOW2019</a>
                  is coming up.</li>
                <li> Switch-over of the farm to 100% Slurm should be
                  announced soon.</li>
                <li> SciComp is interested in feedback on the Slurm
                  experience.</li>
              </ol>
              <h3><span class="mw-headline"
                  id="Discussion.2Fproposal_for_making_transition_to_G4">Discussion/proposal
                  for making transition to G4</span></h3>
              <p>We discussed the <a rel="nofollow" class="external
                  text"
href="https://mailman.jlab.org/pipermail/halld-offline/2019-March/003543.html">proposal
                  Matt Shepherd sent to the Offline Email List</a>. In
                essence it proposes that the collaboration support HDG3
                for data up to including Spring 2017, and support HDG4
                for Spring 2018 data and beyond. This will save having
                to tune both versions to match all of our data sets.
              </p>
              <ul>
                <li> Mark I. thought that the proposal was reasonable.</li>
                <li> Justin also thought the proposal was reasonable.</li>
                <li> Eugene objected to the mechanism where mcsmear
                  bombs when using smearing constants inconsistent with
                  the version of Geant used.
                  <ul>
                    <li> Mark I. thought that bombing has the advantage
                      of preventing the naive user from making mistakes.</li>
                    <li> Thomas pointed out that MCwrapper would have to
                      pass errors of this sort back to the user.</li>
                  </ul>
                </li>
                <li> There was a question about whether the HDDM output
                  from HDG3 and HDG4 records information about which
                  program produced it. No one knew for sure.</li>
                <li> Eugene gave an alternate proposal: he feels
                  strongly that all results, including those that we
                  have completed, and in particular the J/ψ analysis,
                  should be analyzed using both HDG3 and HDG4. We should
                  not go forward and adopt HDG4 as the default until
                  these comparisons with full physics analyses are done
                  and differences understood.</li>
                <li> Justin proposed a compromise via modification of
                  Matt's proposal:
                  <ol>
                    <li> Drop the wording about which versions of HDG3/4
                      are supported for specific data sets.</li>
                    <li> Add that going forward HDG4 is the default.
                      Comparative studies between HDG3 and HDG4 are not
                      discouraged and the ability to do such studies
                      should be maintained, but HDG4 is the default
                      choice.</li>
                  </ol>
                </li>
                <li> Mark D. thought that tuning both versions is not
                  really twice the work. Once a system is worked out for
                  one, it can be applied to the other.</li>
              </ul>
              <p>In the end Mark I. did not think that we had reached a
                consensus on any of the proposals. He agreed to come up
                with compromise wording and send it to the group. Thomas
                pointed out that we can discuss this further at the
                HDGeant4 meeting.
              </p>
              <p>[Added in press: after reviewing the discussion
                carefully, Mark was not able to come up with a proposal
                about which, in his judgment, a consensus could be
                formed. There is not enough middle ground.] </p>
              <hr>
              <h3><span class="mw-headline"
                  id="Review_of_recent_issues_and_pull_requests">Review
                  of recent issues and pull requests</span></h3>
              <ul>
                <li> David found and fixed a problem with the use of
                  mutex in the CCAL monitoring code. It will come in as
                  a future pull request.</li>
                <li> Sean suspects there are some deep pathologies in
                  the analysis library, causing memory leaks and code
                  crashes. This came up in the discussion of <a
                    rel="nofollow" class="external text"
                    href="https://github.com/JeffersonLab/halld_recon/issues/111">halld_recon
                    issue #111</a>.</li>
                <li> We noted that Dmitry has acknowledged Sean's report
                  of a memory leak in CCDB. See <a rel="nofollow"
                    class="external text"
                    href="https://github.com/JeffersonLab/ccdb/issues/61">ccdb
                    issue #61</a>.</li>
                <li> Eugene made a feature request for a variable length
                  list of floats for user-defined uses in the HDDM
                  generated event format. See <a rel="nofollow"
                    class="external text"
                    href="https://github.com/JeffersonLab/halld_sim/issues/44">halld_sim
                    issue #44</a></li>
                <li> Eugene make a request for development of an C++
                  interface to the HDDM event format to hide details of
                  handling the HDDM format from the event-generator
                  developer. See <a rel="nofollow" class="external
                    text"
                    href="https://github.com/JeffersonLab/halld_sim/issues/43">halld_sim
                    issue #43</a>.</li>
              </ul>
            </div>
            <div class="printfooter">
              Retrieved from "<a dir="ltr"
href="https://halldweb.jlab.org/wiki/index.php?title=GlueX_Software_Meeting,_March_5,_2019&oldid=91697">https://halldweb.jlab.org/wiki/index.php?title=GlueX_Software_Meeting,_March_5,_2019&oldid=91697</a>"</div>
          </div>
        </div>
      </div>
      <div id="footer" role="contentinfo">
        <ul id="f-list">
          <li id="lastmod"> This page was last modified on 6 March 2019,
            at 17:07.</li>
        </ul>
      </div>
    </div>
  </body>
</html>