<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,_February_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">GlueX Software Meeting Minutes, February 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> Curtis Meyer, Naomi Jarvis</li>
                <li> <b> JLab: </b> Shankar Adhikari, Alexander
                  Austregesilo, Thomas Britton, Sean Dobbs, Ashley
                  Ernst, Stuart Fegan, Mark Ito (chair), David Lawrence,
                  Simon Taylor, Beni Zihlmann</li>
                <li> <b> UConn: </b> Richard Jones</li>
                <li> <b> W&M: </b> Justin Stevens</li>
              </ul>
              <p>There is a <a rel="nofollow" class="external text"
                  href="https://bluejeans.com/s/nXzNt/">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> <b><a rel="nofollow" class="external text"
href="https://mailman.jlab.org/pipermail/halld-offline/2019-February/003511.html">New
                      version set: version_4.1.0.xml</a></b>. Mark
                  reviewed the new releases in this version set. There
                  followed a discussion of the utility of the new dbg
                  and opt builds. Mark pointed out the advantages as
                  outlined in <a rel="nofollow" class="external text"
href="https://docs.google.com/presentation/d/1DpaLMUnviTgtJ_SRT6iZcaORcKeKNOdAxzic-LlFLPY/edit?usp=sharing">his
                    talk from the January 8 Software Meeting</a>. The
                  sense of the group was that the combination of debug
                  symbols and optimization is too valuable to forego,
                  that even in production where optimization is
                  essential, seeing the added information from debug
                  symbols is important when the code crashes. If this
                  combo is built, there is little use for the dbg and
                  opt versions. dbg, although it has better behavior
                  with the debugger, uses as much disk space as combo
                  but runs slowly. opt, although it is a factor of 10
                  smaller than combo, does not have the debug symbols.
                  So despite having the dbg/opt plan endorsed on January
                  8, we will go back to combo-only builds.</li>
                <li> <b>Moving version set files to new repository: <a
                      rel="nofollow" class="external text"
                      href="https://github.com/JeffersonLab/halld_versions">halld_versions</a></b>.
                  Mark will be moving the location of the version.xml
                  files from their current location in the "dist"
                  directory of /group/halld/www/halldweb/html to a new
                  "halld_versions" directory. This is to facilitate
                  export and update of builds outside of the JLab CUE.
                  <ul>
                    <li> The default version set has always referred to
                      the latest tagged build of each package. Sean
                      wondered if rather it should point to the last
                      versions used in reconstruction. Beni and Mark
                      thought that the current practice was best. New
                      versions come out more frequently than
                      reconstruction launches and if the corresponding
                      version set is desired, it is available.</li>
                  </ul>
                </li>
                <li> <b><a rel="nofollow" class="external text"
                      href="https://mailman.jlab.org/pipermail/cuga/2019-February/002212.html">HOW2019
                      computing workshop: OSG All-Hands + WLCG/HSF:
                      March 18-22</a> (<a rel="nofollow" class="external
                      text"
                      href="https://indico.cern.ch/event/759388/timetable/#20190318">Indico
                      site</a>)</b>. HOW refers to "High energy software
                  foundation," "Open science grid," and "Worldwide LHC
                  computing grid." Richard described the upcoming
                  meeting at JLab. We will prepare a contribution to the
                  session at 2:00 on Monday afternoon, "Input from
                  communities/experiments: Input from other
                  experiments."</li>
                <li> <b><a
                      href="https://halldweb.jlab.org/wiki/images/c/c7/Capture.PNG"
                      class="internal" title="Capture.PNG">Slack</a></b>.
                  Mark reported that he created a new channel, #halld,
                  on the <a rel="nofollow" class="external text"
                    href="https://jlab12gev.slack.com">Slack workspace
                    jlab12gev</a>. <a rel="nofollow" class="external
                    text"
                    href="https://en.wikipedia.org/wiki/Slack_(software)">Slack</a>
                  is a modern chat/messaging application oriented toward
                  the enterprise. Follow <a rel="nofollow"
                    class="external text"
                    href="https://jlab12gev.slack.com/signup">this link</a>
                  to join.</li>
              </ol>
              <h3><span class="mw-headline"
                  id="Review_of_minutes_from_the_January_22_meeting">Review
                  of minutes from the January 22 meeting</span></h3>
              <p>We went over <a
href="https://halldweb.jlab.org/wiki/index.php/GlueX_Software_Meeting,_January_22,_2019#Minutes"
                  title="GlueX Software Meeting, January 22, 2019">the
                  minutes</a>.
              </p>
              <ul>
                <li> The confusion over HDDS geometry has been resolved
                  since the last meeting. It remains to create a
                  repository-based build of the pre-DIRC geometry so we
                  can have a version of halld_sim with consistent builds
                  of hdgeant and mcsmear.</li>
                <li> David reported that NERSC has realized that our
                  typical workflow is not at all like what they normally
                  see. Their projects typically run for much longer
                  (days not hours) and with more nodes
                  (multiple-multi-core nodes versus one). They are
                  preparing suggestions on best practices for us.</li>
              </ul>
              <h3><span class="mw-headline"
                  id="Report_from_the_January_29_HDGeant4_Meeting">Report
                  from the January 29 HDGeant4 Meeting</span></h3>
              <p>We went through <a
href="https://halldweb.jlab.org/wiki/index.php/HDGeant4_Meeting,_January_29,_2019#Minutes"
                  title="HDGeant4 Meeting, January 29, 2019">the minutes</a>.
              </p>
              <ul>
                <li> The problem with "100 times slower" execution of
                  hdgeant and hdgeant4, and the simple solution (CKOV=0)
                  was explained at the meeting by Richard. Thomas has
                  incorporated the fix into MCWrapper.</li>
                <li> Thomas has completed the new set of comparison
                  simulations (HDG3 vs. HDG4). Folks are looking at the
                  result now.</li>
                <li> Added in press: Peter Pauli has done an extensive
                  high-level comparison of HDG3 vs. HDG4. More on that
                  to come.</li>
              </ul>
              <h3><span class="mw-headline"
                  id="Report_from_the_SciComp_Meeting_on_January_31">Report
                  from the SciComp Meeting on January 31</span></h3>
              <p>We reviewed <a
href="https://halldweb.jlab.org/wiki/index.php/Report_from_the_SciComp_Meeting_on_January_31"
                  title="Report from the SciComp Meeting on January 31">Sandy
                  Philpott's notes from the meeting</a>. Highlights:
              </p>
              <ul>
                <li> Two more ifarm nodes will be deployed soon.</li>
                <li> Theory jobs on the farm will be pre-emptable by
                  production accounts, not by all accounts.</li>
                <li> Slurm is almost ready to go into production
                  replacing PBS/Maui.</li>
                <li> More Lustre disk space is coming.</li>
                <li> More work disk space is not coming.</li>
              </ul>
              <h3><span class="mw-headline" id="XROOTD_and_GlueX">XROOTD
                  and GlueX</span></h3>
              <p>With help from Sean to get going, Thomas has had some
                success testing XROOTD. He and Kurt Strosahl got a
                server running on scosg16 (the OSG submit host) and
                Thomas was able to run Monte Carlo on his desktop,
                streaming the random trigger data from scosg16. The
                initial attempt at doing the same on the grid did not
                work, but note that the attempt-on-the-grid count = 1
                right now. We recalled that Richard ran a
                proof-of-principle exercise several months ago using a
                server he stood up at UConn.
              </p>
              <h3><span class="mw-headline"
                  id="ML_Monitoring_and_ML_Tracking">ML Monitoring and
                  ML Tracking</span></h3>
              <p>Thomas noted that there is a push from Management to
                explore machine learning at the Lab. Some weeks ago he
                and Dmitry Romanov gave talks at the ODU-sponsored
                Machine Learning Fest. He showed <a
href="https://halldweb.jlab.org/wiki/images/2/2d/20190205_ML_projects.pdf"
                  class="internal" title="20190205 ML projects.pdf">a
                  slide</a> demonstrating his work in classifying BCAL
                occupancy plots, generated by the online monitoring, as
                either "good" or "bad." He is using <a rel="nofollow"
                  class="external text"
                  href="https://www.tensorflow.org/guide/keras">Keras/TensorFlow</a>.
              </p>
              <p>David has been doing work using ML to classify tracks
                based on the raw hits alone[?]. The 39 slides following
                Thomas's one in the pdf file linked above describe his
                work. We ran short of time; he will present this
                material at the next Track Meeting on Thursday.
              </p>
              <h3><span class="mw-headline"
                  id="Review_of_recent_issues_and_pull_requests">Review
                  of recent issues and pull requests</span></h3>
              <p>Sean drew our attention to the halld_recon issue he
                opened today, <a rel="nofollow" class="external text"
                  href="https://github.com/JeffersonLab/halld_recon/issues/95">Crashes
                  with analysis library when adding multiple similar
                  reactions #95</a>. Alex has some ideas about the
                cause.
              </p>
              <h3><span class="mw-headline"
                  id="Code_Readiness_for_Reconstruction_of_Spring_2018">Code
                  Readiness for Reconstruction of Spring 2018</span></h3>
              <p>Sean reported on some bugs that he and Simon are
                working on that need fixing before we can go ahead.
                Simon has also submitted <a rel="nofollow"
                  class="external text"
                  href="https://github.com/JeffersonLab/halld_recon/pull/94">a
                  pull request</a> with several changes to tracking.
                Sean will report back when we are ready to proceed.
              </p>
              <p>[Added in press: Simon closed his pull request. He will
                resubmit the bug-fix-like changes as a separate pull
                request from the more fundamental tracking changes. The
                former will definitely be included in the reconstruction
                launch.]
              </p>
            </div>
            <div class="printfooter">
              Retrieved from "<a dir="ltr"
href="https://halldweb.jlab.org/wiki/index.php?title=GlueX_Software_Meeting,_February_5,_2019&oldid=91132">https://halldweb.jlab.org/wiki/index.php?title=GlueX_Software_Meeting,_February_5,_2019&oldid=91132</a>"</div>
          </div>
        </div>
      </div>
      <div id="footer" role="contentinfo">
        <ul id="f-list">
          <li id="lastmod"> This page was last modified on 5 February
            2019, at 19:41.</li>
        </ul>
      </div>
    </div>
  </body>
</html>