George: issue in Alma Pol Helpers Recipes: Check in this morning
Issue with libraries reference
Dirk's Issue for plot MS: Pam to merge across.
Sandra, small fix on concat to fix parallel regression.
Turning on the C++ tests
Files starting with d are demonstrator codes
Files with t are the tests
Purpose built MSs
Kumar writing functions to build MS
Namespace will be casa::test
Test fixtures go "close" to the definition of the code.
Get rid of warnings in the compile
Start making the transition away from boost.
We had a telecon with NAOJ today. Takeshi has fixed the remaining problems with releasing GIL in the SD pipeline. He reported a speed-up ratio of about 4.1 in both baseline fitting stages, which is better than the expected, mostly probably due to several performance improvements in context and casalog handling.
We are working on the documentation about parallelization for the cookbook. The main issue is that the cookbook recommends the user to look at the alma-m100-hpc script as an example on how to run an analysis in parallel, but this script has been failing for a few years. I have fixed a few things in importasdm (MMS section) and in virtualconcat and now the scripts does not fail but the statistic comparisons at the end still fail. TBV.
Akeem and Sandra to get M-100 tests (HPC version) into accepted tests.
Documentation (Carried to next week).
C++ test - namespace and repository of helpful testing functions
Development and testing of standard ALMA imaging recipe for pipeline C3R3 release
Primary beam corrections, masks, line finding
Flagging and calibration heuristics testing for C3R3 release
Changed the order of the creation of an MMS inside importasdm so that the .flagversions are properly created.
Fixed virtualconcat so that SYSCAL and SYSPOWER are also created as symbolic links in the output MMS.
The call to clean inside alma-m100=analysis-hpc-regression.py needs to be translated to tclean, instead of pclean. Who can do this?
Working on the HPC documentation for the cookbook.
Fixed interactive flagging bug where only part of user's flags were saved back to MS (no JIRA ticket)
Discovered virtual model absolute path issue with ms copied from /lustre/naasc; Kumar put in a fix
CAS-7823 RTT - check more often if user canceled cache-loading
CAS-7830 RTT - changed progress widget to cancel if user clicks "Cancel" or "X"
CAS-7831 under test - tester said plotting default virtual model was not fixed but it worked for me with her test data set
Worked on restarting plotms when it stops without restarting casapy session; just need to handle logging
Writing a VisModelData test for comparison of real and virtual model in C++
fixing an issue on msuvbin where some baselines get super high values
Looked at an issue in tclean with restoringbeam='common'
Fixed an issue with virtual model when the ms is moved or copied
Justo González Villalba
Attended ERIS interferometry school
Refactor DDI/SPW/Pol methods of mstransform to avoid inconsistencies reported in CAS-7814/CAS-7841
Completed user testing of OTF calibration via mstransform
Fixed sign issue reported in OTF phase shifting applied via plotms
Implemented a generic weight propagation method in mstransform applicable to all linear combination kernels.
Misc meeting about pipeline timeline etc
CAS-7676: Fitting in non-optical mode does not work when rest freq is not set (investigated & assigned to Dave)
CAS-7681: Unit is not properly displayed in spectral profiler (ready to test)
Three days running selenium tests and fixing associated test & code problems.
Helped Alex track down deployment problem.
Work on supporting images with nontraditional axes.
(CAS-7599) imfit deconvolved peak flux unit: Reassigned to Juergen for requirements gathering.
(CAS-6979) imfit should have an option to report the size of a uniform disk (rather than a Gaussian) that would be consistent with the image fitted size when convolved with the beam: reassigned to Jeff for management of algorithm development.
(CAS-7778) msmd should have a method to retrieve the effective bandwidth or the effective resolution of the channels rather than only their formal bandwidth: RTT, implementation complete.
(CAS-7869) misspelling of "parameter" in specfit: Fixed.
(CAS-3366) listfile parameter for imstat: RTT, modified output as requested.
(CAS-7637) msmd should have methods to retrieve REST_FREQUENCY and TRANSITION from the SOURCE table: RTT, completed implementation.
(CAS-7677) immoments crashes when calculating moment 0: Fixed, images with rest frequency=0 now supported.
(CAS-7678) calculation of moment 1 leads to map with only NaNs and Masked pixels: Fixed, images with rest frequency=0 now supported.
Received new macbook. Unfortunately it has only a single USB port on the monitor port side which means both connectors for my KVM mouse and keyboard cannot be simultaneously connected, so requested a small USB hub.
(CAS-7850) immoments needs to support images with 0 rest frequency: Fixed, images with rest frequency=0 now supported.
Made image supplied by Dirk available to Susan for her investigation of a viewer ticket regarding the same image.
Replied to documentation discussion voicing my opinion that taking time to design the structure of developer documentation is required for having a useful and maintainable documentation system. Kumar's later response seemed to show support for this view.
(CAS-6896) immath should give control over file used for header generation: Refactored the ImageAnalysis::imagecalc() bits into their own class. Pondered why there are independent implementations for ia.imagecalc() and ia.calc() when the only real functional difference seems to be that imagecalc() returns an image tool while calc() overwrites the current image tool. Some design choices made years ago in the image analysis module continue to mystify me.
(CAS-4663) specfit components/prompt/inf: RTT. Implemented what I hope is the last requirement for this ticket, namely each component should have its own solution and/or error image when such is requested, rather than bundling up several components in the same image.
(CAS-5524) import uvfits should do make consistent when getting visibilities in STOKES form: Fixed. importuvfits does not support fits files with Stokes parameters, rather than real correlations (LL, RR, etc), and an exception is now thrown if a user tries to import such a file.
mostly worked on finalising the ephemeris support
CAS-7301 was used to track this (even though the title is misleading): result: resolved but found that ICT-3873 needed to be reopened, PREQ-331 was filed
found CAS-7878 (plotms bug concerning radial velocity problem) and fixed in trunk; asking for permission to get this into release branch
tested and resolved a number of image analysis related tickets
tclean inline docs
many emails about pblimit and T/F vs 1/0 masks + fixed couple of minor bugs
made a makePB recipe for use with tclean for non aprojection gridders; still polishing it.
turned on subset of tclean regression unittests (~50 small ones) to run automatically
fixed tarball casa not starting on Debian based systems (CAS-7196)
fixed random mstransform test failures, caused by uninitialized use in recent mstransform changes
updated casa rpms to include mpicasa
Fixed segfault in cubeimagepartition() + a few (unrelated) bugs for parallel cube tclean (CAS-7862)
Re-tested interactive mask issue in cube tclean but couldn't reproduce (CAS-7669)
Fixed JPLephem_reader() for JPL-Horizons ephemeris data with finer time increments CAS-7872)
Friday NAOJ Meeting
casa pipeline 4.3.1 was installed on the analysis servers at NAOJ.
Resolved CAS-7814 and CAS-7838 (Thanks Justo and Michel!)
Fixed wrong help description on tsdcal
Supported Erik to created MS-based (i.e. non-ASAP) single dish reduction script
release GIL in ASAP and test on pipeline done (CAS-7764)
CAS-7339: Detailed specifications clarified with Takeshi and captured in JIRA comment