Monday Morning Meeting 6th June 2015

  • DIAL-IN NUMBERS & PASSCODES:
  • IP: 192.33.117.12##8110
  • Phone: (434) 817-6524

Attendance

  • Socorro:
  • CV:
  • Garching:
  • SCO:

News / Meetings / Visitors

  • Summer School this week.
  • Peter Teuben and Lee Mundy: June 6-8 ADMIT
  • Ilse and Des after the summer school (VLBI)
  • Kelly Sharp June 13 (Plone & Documentation)

Build, Release

  • update to devtoolset-4 for el7?
  • Bamboo CI builds

Verification Testing

  • Converting accepted tests to groovy/gradle based setup, this was due to CASA not exiting properly in the old perl based system
  • Osiris (which is where casaguides were being tested) was split from the rest of the cluster, and promptly had what is believed to be a hardware failure
    • Still not working
    • Running casaguides manually on the cluster
  • Still updating the spreadsheet, should have it all updated, and tests grouped together by mid-June

Validation Testing

Architecture

  • Warnings Summary -- All modules but Alma, synthesis and singledish are green. Still need to devise a way to monitor warnings for all supported platforms. That will probably require a nightly (?) job that does a clean build.
  • Google Test now running on OSX. Three tests are currently disabled on OSX because they are failing: LineFinger_Gtest, LineFindingUtils_Gtest, and tRegriddingTVI; please see if these can be fixed and reenabled. Disabling was accomplished by adding the argument NOT_ON_APPLE to the macro call that defines the test in CMakeLists.txt. Remove that argument to reenable the test under OSX.

HPC

Development

Pipeline

  • C4R2 development plan for pipeline F2F meeting, 1 month to go
  • Pipeline HLTAU results
  • Issues, performance, OBJECT keyword, NaN problems seem to be fixed

AOB

Developer Reports

Monday Meeting
  • Sanjay Bhatnagar
  • Sandra Castro
  • Lindsey Davis
    • Reworked hif_uvcontfit to generate one caltable per source
    • Updated the hif_uvcontfit web log page to display the per MS TOPO spw / frequency range selections
    • Updated hifa_fluxcalflag to flag all SSO emission line regions even if the entire spw is flagged
    • Updated hifa_fluxcalflag web log page to display a warning if the reference spw map is non-trival
    • Started on hifa_spwphaseup error handling and web log improvements
    • Email discussion on pipeline tclean performance, OBJECT keyword handling, issues
    • Started F2F meeting / C4R2 jira ticket triage
  • Pam Ford
    • CAS-8487 (avgchannel in plotms cannot handle value of 1e10) - responded to tester's comments, now Resolved
    • CAS-8381 (Add plotcal poln selection to plotms) - input required
      • worked through 3C286 polarization calibration in casaguides to create test cal tables
      • cannot determine polarization basis from cal table keywords ("unknown"), asked for input
    • updated test_plotms and test_mstool
  • Kumar Golap
  • Justo Gonz├ílez Villalba
    • Tested latest XCLASS release (1.1.8) vs CASA 4.6
    • Completed cvel TVI includign corresponding GTests
    • Now working on SPW combination and UVContSub TVIs.
  • Jim Jacobs
    • Closed out cas-8438: When adding scratch columns only a single default tile-size went to the new column. In this case the WVR channel became the default tile size (only one correlation and 4 channels) which required a huge number of tiles to cover the main floatdata column which had 4080 channels. Ger modified the table system to add the ability to clone a column (while allowing element type change) and to preserve the tiling of hypercubes when copying or filling columns. Now the scratch column will pretty much match the data column structure.
  • Jeff Kern
  • Susan Loveland
    • Synthesis Imaging Workshop
    • CARTA:
      • Ds9 region parsing (Ville is helping with MAC build problem)
      • Context and zoom view windows for image.
  • David Mehringer
    • imstat slow on stacked images (from parallel clean). WIP. Investigated and found that tclean is configuring these images to be open and closed after each lattice iteration. In the image provided, there are four concatenated images and it takes 16,000 iterations to step through the entire image, so my naive count is that there are 64,000 open/close cycles for the underlying images plus as many for the underlying masks, so 128k open/close cycles. Given that it appears to be a requirement for tclean to write ImageConcat images configured in this way, and that most image analysis tasks must use LatticeIterators to step through images, I don't think there is much that can be done to improve performance here; these processes will be very much I/O bound.
    • (CAS-8710) immath cannot handle files produeced in parallel tclean. WIP. Ger says this should work, but it does not work on the supplied image, so I've punted to him.
    • Posed questions about the fits test failures on RHEL 7. Apparently they are due to wcslib being upgraded.
    • (CAS-8713) create msmd.spwsforscans. RTT. implemented.
    • (CAS-8681) create msmd.statesforscans(). RTT. implemented.
    • (CAS-8632) add msmd.spwsforfields(). RTT. implemented.
    • Work to enable several ia tool methods to write image history.
  • George Moellenbrock
  • Dirk Petry
  • Martin Pokorny
  • Urvashi Rao
  • Darrell Schiebel
    • Boost cleanup in bnmin1
    • improve linux packaging of Xvfb
    • setup NFS for serving data from Apple/Fedora host to virtual OSX 10.11 guest
    • work getting Structure101 to use Understand udb static analysis database
    • add spectral fixed precision option to the viewer ( cas-8633)
    • add contour line width as an option to imview(...) ( cas-8366)
    • unifying wcslib on RHEL6 & RHEL7 around 5.15 (pgsbox & runpath)
      • new wcslib RPMs will be added to RHEL6 yum repository June 6th
  • Ville Suoranta
    • Updated run_casa to support --pipeline option
    • Carta builds: RH6 builds now. Learned that the existing package doesn't have python packaged properly. OS X and RH6 instructions in pretty good shape. RH7 needs work
    • Initial CI builds on Bamboo
    • Initial packages+tests (RH7 and OSX) on Bamboo
    • Investigating Jenkins build issues
    • Worked with Jim on OSX Google tests
  • Julian Taylor
    • setup local el7 regressions, fixed huge memory leak of plotms causing failures
    • debugged binary test failures since el 6.8 upgrade of build machines, fixed by Darrell
    • CAS-8618: copy ephemeris table in split if reindex=False
    • a couple casacore code reviews
    • fixing broken agentflagger plots on el7, does not crash anymore but they are still empty ._.
  • Takahiro Tsutsumi
    • Looked into a couple of parallel cube tclean issues reported by Juergen
      • Fixed the image concatenation error for restart = T (CAS-8702)
      • Investigated parallel cube tclean failure when a startmodel specified
    • Got update request for the broken polynomial coefficients for 1934-638 (ATCA flux calibrator) by Mark Wieringa (CAS-8711). Proposed to create a new standard for that rather than updating the existing standard(s) (the source is not in Perley-Butler 2013).

Friday NAOJ Meeting
  • Kanako Sugimoto
    • a couple of fixes to sdimaging and tsdfit
  • Wataru Kawasaki
  • Masaya Kuniyoshi
  • Takeshi Nakazato
    • resolved CAS-8419
    • CAS-8155 comments from tester
    • CAS-8416 implemented worker class for gain calibration based on script from Stephen White
  • Renaud Miel
    • CAS-8082 PlotMS Header
      • PlotMS code + doc review, Qt doc reading
      • PlotMS practice / multi-plots, Qt practice / small app compile + run

-- KanaSugimoto - 2016-06-02
Topic revision: r23 - 2016-06-06, DaveMehringer
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback