Warnings: Removed all warnings from msvis module and marked module in code/CMakeLists.txt as WarningsAsErrors. From now on any warnings in this area will be treated as errors. Other modules can be similarly marked when warning-free. Example:
Unfortunately ran afoul of something on Jenkins which issues warnings on a command line which runs without warnings on my box. Further investigation required.
Google Test: Still fighting to get google test support into build system with Ville's help. It's very close, but Jenkins is using a version of Jenkins older than what is on the RHEL 6 developer machines and it is behaving differently. During make the Google
Testing Guide: Updated testing guide to better outline plan going forward for unit and module tests (module tests are limited scope but longer running than unit tests).
While running performance tests on the pipeline, run into a tclean problem when setting up the image shape (CAS-7943). Stewart confirmed the failure on his side. Tak fixed the issue, I've tested it, but now the pipeline stops later in stage 22. Stewart will verify it. These failures only happen when running in parallel.
Working on presentation for CUC.
Working on presentation for CUC.
Found an error in test_mstransform_mms, which fails because of a mismatched row in SPECTRAL_WINDOW table. Justo fixed it in r34972.
Other work unlelated to CASA.
CAS-7934 RTT (text box for setting grid line width is too narrow on OSX 10.10)
CAS-7933 RTT (precision for seconds in timestamps should be 0.1ms in plotms Locate)
CAS-7946 (improve test_plotms): refactored, added tests, grouped together similar tests, all test MSes in casa data repo. TBD: set image resolution for better export file testing.
Looked into casapy --noipython switch for helpdesk ticket
Testing msuvbin on some 500 GB of data
Helped Claire and Urvashi debug some crashes and other related bugs in tclean
worked on some simulation for testing purposes
Justo González Villalba
Minor mstransform fixes related with HPC context
Preparing a parallel test for the OTF calibration in mstransform
Working on an improvement of phase shifting for 4.6, in order to upgrade TVI phase shift to match fixvis
CASA Users Committee
CARTA: Worked on bugs and viewing images with different axes.
Worked on PEP
(CAS-7937) imsmooth produces strange warning message: RTT. Removed undersampling warning message in case where input image has no beam.
Read some of Eduardo's report.
Responded to Takeshi N. Re: tb.putcell()/getcell()
Tried to write a C++ test using google test in code/ but didn't work. Apparently, the google test infrastructure has not been pushed out to NRAO developer (or at least my) machines yet.
(CAS-5346) Reviving ia.histograms(): RTT. Modified as requested to also return standard deviation and mean of each histogram. Also did a lot of refactoring to pull the high level histograms related code out of ImageAnalysis into its own class which is derived from ImageTask.
(Re)Took ownership of some image analysis documentation related JIRA tickets from Mark.
(CAS-6744) Correction to quartile definition in imstat/ia.statistics help: Resolved. Confirmed that I had done this as part of the stats framework implementation last winter.
(CAS-4663) specfit components/prompt/inf: RTT. Reimplemented inf/-inf masking using isInf(Float) rather than isInf(Double) which resolved the issue.
(CAS-7704) Image analysis task should have 'chans' and 'stokes' to override par.region specification: RTT. Updated the related documentation to match the new rules I had implemented, but not committed, in July and committed the implementation + documentation.
Fixed imstat test that my commits for CAS-7704 broke (I'm surprised that particular piece of code compiled, given a pointer was passed where a String should have been).
(CAS-6785) Please update componentlist.addcomponent Documentation: RTT. Updated docs as requested.
(CAS-7959) Listobs Segfault on MS: This dataset, which took 16 hours to copy from Socorro to Charlottesville, is invalid. There should be (better) checking in the MeasurementSet constructor to ensure the MS it is given is actually valid, otherwise, when this supposedly valid object is handed to other classes for processing, chaos (often in the form of segfaults) will ensue.
Mainly explored mechanisms for generating calibration errors for tests.
Darrell Schiebel worked on the following
data update (casa-rsync.nrao.edu) for linux distributions
removing boost::foreach dependency
tracking C++ include dependencies in Java
restored linux binary 'bin' directory
changed a line in wvrgcal (remove unnecessary function call) -> 30% faster
(took two days off)
Fixed a bug in parallel cube tclean (CAS-7943)
Some science time: discussions on ALMA Cycle 2 data reduction path with the PI
Friday NAOJ Meeting
vacation until 13th Oct.
some investigation on CAS-2799
weight handling for gaussian smoothing CAS-7874
CAS-7339: Mixed Python/C++ debugging/stepping of Casa unit test (test_tsdcal.test_ps05) in Eclipse GUI