CASA Users Committee: 26,27 Sept. In Socorro. Will discuss attendance with individual team members soon.
No Meeting next two weeks (June 27 (Jeff Travel), July 4: Holiday)
New CASA Stakeholder Meeting
Lewis Ball will be the new AD of SSR. Starts in Sept or Oct. Lewis will probably not be responsible for SRDP.
Build, Release
test_tsdbaseline failing consistently on RHEL7. Setting up RHEL7 host at NAOJ. Hopefully this will be fixed soon.
Verification Testing
Akeem working w/ Pam on ADMIT. Can build it, but is not automated yet. Also working on topical VLA CASA Guides. Working on running CASA Guides on OS X.
Pipeline testing:
There are two test cases that we will try to support in the automated system.
A fixed version of the pipeline will run on the latest stable of CASA. This will identify backwards-compatibility changes that are being introduced. As we approach a release, the fixed version will approach the final release version.
A fixed version of CASA will build against the bleeding edge of the pipeline. This will support the pipeline development team.
Looking into a schedule that puts pipeline on a yearly release.
Currently there is one regression test that Akeem runs on the pipeline.
Process underway to hire a new member of test team.
Integrated Crash Reporter with PlotMs and the Viewer.
Integrating Crash Reporter with the build system. It will require aa cmake flag to enable its incusion in the build and should be used only for non-development builds. The integartion also includes producing the symbol extracts of the executables and libraries. Logic is working for that in the code project. Since the CasaCore build system is shared, handling CasaCore is TBD at this point.
HPC
pipeline run with tclean auto-chanchunk selection successful while last time it always failed, but the selection always choose chanchunk=1, has there been some memory usage bugfix in the last few weeks?
main tclean workhorse function sectggridd seems to have some optimization potential
Development
Python 3
Discussion has been postponed for a long time. Starting to think about when to deal with this. This is a big change for developers and users. We may release casa 5 as python 3, but also release casa 4.8 which is python 2.
Some 3rd-party packages appear to be staying with Python 2.
Some libraries are moving to Python 3 by 2020.
Pipeline
intent selection: code cleanup, improved subscan intents handling for ALMA data sets
calibration: followup on potential combine='spw' solves and multi-tuning ASDM issues
imaging: improvement in noisy edge channel handling, auto-boxing support discussions
web log: navigation improvements
AQUA report generation: worked on calibration summary started
CASA / pipeline issues
data selection issues in ms.msselect, immoments
continuum finding in noise dominated data
tclean divergence in noisy data
AOB
Developer Reports
Monday Meeting
Sanjay Bhatnagar
Sandra Castro
Lindsey Davis
AQUA report generator development, calibration summary
pipeline intent selection discussions
ALMA science data product flow discussions OT->APDM->ASDM->pipeline->AQUA
pipeline infrastructure deployment discussions with SSA group
Pam Ford
CAS-8612 Resolved (weight/sigmaSpectrum with spw selection) - Not a Bug. Discussed with Jim and tested some more. Differences in copy ctor vs. assignment for the returned cubes.
CAS-7817 RTT (Bug while iterating over SPW) in light of CAS-8612 put in a bug fix.
ADMIT on RH7 (success) and OSX 10.10 (fail)
More reading on calibration and cal tables
Bought a house
Kumar Golap
Justo González Villalba
Implemented first version of UVContSubTVI following uvcontsub3 approach (polynomial fitting)
Made some basic optimizations related with calculation of frequency pows
Added some flexibility in the low level transform methods to allow "want_cont" option as in task_uvcontsub
Pending to optimize usage of polynomial fitting classes, possibly introducing support for Complex vs Float fits.
Pending GTests vs uvcontsub3
Provide support for uvcontsub issues reported from VLA/Cosmos field survey
Jim Jacobs
Ger is going to look at the column keywords on create problem (cas-8427?).
Debugged problem with a Carta build. Tried chasing down an environmental problem but turns out the mistake was assuming that the committed code actually built; it was simple once I stopped believing that.
Jeff Kern
ALMA Lifecycle
Job Ad for CASA Liaison
Documentation Meeting
Susan Loveland
Finished first draft of Image/Context Views (Kumar provided help with calculation of angles between coordinate axes)
Started work on integrating Pavol's Cyber SKA spectral fitting code into CARTA (Jim provided help with build problem in Pavol's makefile file regarding the gsl library).
David Mehringer
(CAS-8741) exportuvfits seems to not set CRVAL of FREQ axis right: RTT. Fixed
(CAS-8477) exportuvfits not honoring multiple spectral windows in FEED table: Added check for varying receptor angles.
(CAS-8725) imhead(mode='put') should write history to image: RTT. Implemented
(CAS-8743) remove WSRT specific code from MSFitsOutput: Resolved. Removed dead code.
Contemplated how to parallelize imstat/ia.statistics()
George Moellenbrock
Some JIVE/VLBI followup, including: FITS-IDI ANTENNA.OFFSET discussion (should be handled by correlators and merely recorded in MS, not used in calibration); check-in of FringeJones class to support fringe-fitting solve/apply (tool-driven infrastructure recognizes it, and it works mechanically!)
Determined that Lindsey's combine='spw' problem (that applycal yields apparently uncalibrated data) is in fact due to a phase jump in one spw that makes use of combine='spw' a bad idea (spws are not consistently coherent within the dataset).
Misc. personal build wrestling (NB: "NOT_ON_APPLE" was turning off all tests after the first specification of this keyword in a CMakeLists.txt file; Takeshi seems to have independently discovered this and supplied a fix on June 17.)
Continued staging various components of VI2 in the calibration solve....
Dirk Petry
Martin Pokorny
Further IO testing to characterize effect of tile shape
(VLA: subarrays in CBE)
Urvashi Rao
Made progress on getting 'makepb' and 'pbcor' into the refactored imager, using whatever beam model the vpmanager has.
Minor fixes and user support
Darrell Schiebel
cleanup of CV host 'warp' in preparation for RHEL6 upgrade
changed casa log file from a 'casapy' prefix to a 'casa' prefix
started building python3 RPM
Ville Suoranta
Julian Taylor
implemented heuristics for selection chanchunks in tclean which defines memory usage
Takahiro Tsutsumi
resumed to work on auto masking in tclean
~2 days on science: for reading/commenting on drafts of two papers
Friday NAOJ Meeting
Kanako Sugimoto
reorganization of 4.7 targets with Erik
discussion of failures of checking out ASAP via http protocol ... switched to https
investigation of google test failures ... sent note to Jim. Need to set CASAPATH for ctest.
confirmed Jim's fix to tile shape of scratch columns ( CAS-8438)
processed the latest B10 TP data to test side band separation using image. This work will be discarded from 4.7 target I'll hold this work.
Wataru Kawasaki
Masaya Kuniyoshi
Takeshi Nakazato
working on CAS-8416
part of this development depends on George's work which is to migrate VI/VB dependent code to VI/VB2
reported some problem on 'variable' mode of tsdbaseline