Monday Morning Meeting 1 Feb 2016

* NOTE: Special Dial in this week!
  • DIAL-IN NUMBERS & PASSCODES:
  • IP: 192.33.17.12##8110
  • Phone: (434) 817-6524

Attendance

  • Socorro: Jeff, Lindsey, Susan, Urvashi, Kumar, Pam, Martin, Tak, Juergen, George
  • CV: Akeem, Andy, Dave, Jen, Darrell, ViIlle
  • Garching: Justo, Sandra
  • SCO:

News / Meetings / Visitors

Build, Release

  • 4.5.1-6
  • 4.6: Missing the code freeze
    • SetJy Astroid Model (waiting for Butler)
    • VI2 Integration (Still behind) This week
    • Flagging Generalized Pre-averaging (checked in)
    • MPI for CASA (Justo- Couple of days)
    • Partition by baseline average is in.
    • Cvel Error Message
    • Dave: Small feature for mdmd

Verification Testing

  • Pipeline test is failing due to flux values changing
    • Previous values was 0.716
    • Now it is 0.694
    • Test relative tolerance is currently 1.0e-5
    • absolute tolerance is 1.0e-8
    • Understanding of why the value changed would be ideal, in lieu of that, should the reference value be changed, or tolerances be loosened to get this in the green?
  • High powered machine is now avaliable for casaguide testing
    • casaguides have been run on there, manually, still under development. Run times have been shortened based on testing on other hardware - roughly 1/3 improvement in run time
    • Differences in image comparisons is challenging (as in png comparisons), due to as little as 1 pixel off between plots
  • Added plotms and split to OSX accepted tests

Validation Testing

  • No significant update from last week (JDM "redeployed" last week)
  • CAS-8167?
  • ATM Library has been updated.

Architecture

Warnings

  • We're down to two modules showing warnings on RHEL6: Alma (79) and SingleDish (30).
  • Still need to incorporate logic to watch warnings in the actual Jenkins system and expand coverage beyond RHEL6.

Unit Tests

  • Since the start of the year, only the SingleDish module has increased the percentage of tests it has online.
  • Current status web page.

HPC

Development

Pipeline

  • tclean will support a fixed set of image products option for the pipeline and scripting use cases, thanks Urvashi !
  • development of low SNR gaincal heuristics for ALMA interferometry pipeline, requires combine='spw' and spw mapping, some web log issues
  • efficiency of pipeline domain and callibrary modules improved, but still a ways to go to support VLASS data
  • follow up on imaging issues, apparent sensitivity for pointed mosaics and online binning cases, what to do when continuum detection fails
  • Tsys flagging heuristics improvements, and flagging code refactor
  • plotbandpass plot file name and contents confusion, needs to be patched for CASA 4.5.1

AOB

Developer Reports

Monday Meeting
  • Sanjay Bhatnagar
  • Sandra Castro
  • Lindsey Davis
    • Implemented low SNR gaincal heuristics for ALMA interferometry pipeline, some web log consequences still need to be addressed
    • Followup on C3R4 release issues
  • Pam Ford
    • CAS-8231 (Add feed selection to plotms) RTT - Thanks Jim for merge on Github casacore!
    • CAS-8247 (Create ms tool tests) Resolved and Closed - added test_mstool.py to python test suite
    • CAS-8197 (Plotting field/scan selection) Input Required - Investigated no plot; suspect null selection
    • CAS-8179 (Plotting WtSp) Under Test - Investigated "QObject::moveToThread" error, could not reproduce. Ready to resolve.
    • CAS-8050 (Remove MSSelector from ms tool) - created additional tickets for this project and continued work
    • Reviewed plotms tickets, suggested items for 4.7 release
  • Kumar Golap
    • put in code for automatic wprojplanes in synthesisimager
    • explored reports of virtual model not mapping properly on correlations...not a buug
    • Looking into a report of mosaic behaving differently on different spw's
  • Justo Gonz├ílez Villalba
    • Integrated ChannelAverageTVI in flagging module
    • Enabled and tested generalized (time/chan average) for all visibility flagging agents (tfcrop, rflag, clip)
    • Improve handling of antenna selections re-indexing in mstransform to be based on MSSelection instead of scanning the main table
  • Jim Jacobs
    • Completed prototype integration of crash reporter with casapy. Creates dump files on crash.
    • Have a ticket into NRAO IT to figure out how to get reports back to NRAO from user site. Probably will use "curl" facilities to transmit dump files to web server (write only access required).
    • Still working to rework the VI/VB2 framework to make it easier to chain together transforming VIs.
  • Jeff Kern
  • Susan Loveland
    • CARTA:
      • Now able to generate & display profile using Pavol's algorithm
      • Changing units of profile x- and y-axes
      • Displaying interactive channel line (still need to put in interactive drag)
      • Some infrastructure work for displaying multiple profiles.
  • David Mehringer
    • (CAS-7860) new task-based interface for image moments: Responded to query re: testing.
    • (CAS-7081) After changing RECEPTOR_ANGLE in MS either exportuvfits or importuvfits resets it to zero: RTT. Fixed.
    • (CAS-5846) IMPORTUVFITS in casapy-test failed: IR. Requested data set which exposes the issue.
    • (CAS-109) predefined strings attached to ms. Resolved. Won't fix. Facilitated closing of this nine year old ticket.
    • (CAS-2157) Write unit tests for exportuvfits. Suggested closing of this ticket, as it serves no meaningful purpose, given that it has been open for several years and no substantial work has been done on it. Unit tests are generally written by developers when creating new features or fixing bugs. Unfortunately, this best practice was often not followed during large periods of the project's history. History tells us that, since we do not have infinite resources, we have never been able to provide large chunks of time to people in order to write comprehensive unit tests after the fact, that should have been written at the time when new features were developed. Having open tickets that serve no useful purpose just adds to the noise of such tickets.
    • (CAS-5818) importuvfits sets dish diameters to 0 for 'EVLA' fits file imports. RTT. Fixed. Initially was going to resolve as Won't Fix since AIPS was responsible for writing invalid dish diameter values, but learned that AIPS is purposefully doing this. Since there seems little hope of getting this fixed upstream, I've modified the code to fallback to the standard algorithm if the FITS writer has written invalid dish diameter values.
    • (CAS-1834) exportuvfits creates TY table: IR. It seems a decision by the EVLA people was pending, blocking progress on this ticket, when this ticket was last updated in 2010. Requested the relevant people be pinged for the status of this decision.
    • (CAS-3883) Importuvfits progress meter reaches 100% but task is only 10% complete. WIP. Requested and received requestor's proposed resolution to this issue.
    • (CAS-3660) importuvfits can't handle gaps in Antenna tables. RTT. Fixed.
    • (CAS-8287) CASA image headers should include the systemic velocity of the target to accompany the rest frequency (which it already contains). Resolved. Ticket was mis-assigned to me for coordinates issue, which I reassigned to Dirk. Jeff later resolved the issue.
  • George Moellenbrock
  • Dirk Petry
  • Martin Pokorny
    • Cleaning up issues in visstat2
    • Discussions on integration of MPI C++ code into CASA
  • Urvashi Rao
    • cas-8271 : Gave tclean an option to produce a fixed set of output image to make life a bit easier for scripters...
    • cas-7945 : Enabled a 'restore only' path through the tclean control flow. One could re-restore their images without redoing any imaging or deconvolution.
    • cas-8190 : Enable wprojplanes=-1 through the interface of tclean (to trigger automatic wprojplane calc code from KG)
    • cas-6671 : Changed the behaviour of 'startmodel' to take in full image names (and lists of names for multiterm). This conforms to what clean does too and was in response to users finding the prev interface strange.
    • cas-8066 : tested parallel continuum tclean model writing, found a minor bug in the partitioning routines, updated documentation regarding the state of modelcolumn writes for parallel tclean.
  • Darrell Schiebel
  • Ville Suoranta
  • Julian Taylor
  • Takahiro Tsutsumi
    • Checked in mask interface changes for tclean
    • CAS-7863, new asteroid models support. Most of the modifications were done. Sent the modified code for solar_system_setjy to Bryan for review and comments.
Friday NAOJ Meeting
  • Kanako Sugimoto
    • new modes in initweights ( CAS-7450): done implementing tests. Need proper handling of SIGMA_SPECTRUM
    • DONE intent selection in tsdbaseline ( CAS-8168), tsdfit ( CAS-8169), and tsdsmooth ( CAS-8170): done implementing tests
    • Things to do towards 4.6
      • issue found in antenna and pol selection of sdms tool
      • proper handling of SIGMA_SPECTRUM in initweights
      • Update of ATM library
      • tests of FEED selection ( CAS-7320 implemented by Pam)
  • Wataru Kawasaki
    • CAS-7911 (higher-order interpolation for SD fast-scan data): implementation underway.
    • CAS-7905 (Lorentzian profile fitting in tsdfit): committed a test data.
    • Made patch files to make SD codes successfully built and run with the coming libsakura 2.0. Waiting for the release of CASA 4.5.1...
    • Communicated with Dr. Atsushi Nishimura at NRO as he uses tsdbaseline for NRO-45m dataset.
  • Masaya Kuniyoshi
    • found useflags-related bug.
    • finished with all unit tests regarding visstat2 except for useflags-related unit tests.
  • Takeshi Nakazato
    • working on CAS-7910 scantable to MS filler
  • Renaud Miel
    • CAS-7339 calmode=otf
      • manually verified correctness of calibration results against testcase data with known expected results
      • next steps: verify results on data requiring declination correction, implement input data reference frame conversion, add unit tests

-- JeffKern - 2016-01-26
Topic revision: r21 - 2016-02-01, JeffKern
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