How much work is this: Similar to what is happening with Rest frame
Add frame to Measures, work in the profiler to prevent user from doing anything.
Dirk to follow up with Ger/Wim
Re-working the heuristics of mstransform and split2 when the input is an MMS.
Re-organizing the class used by all parallel tasks to be more modular. Will create a ParallelDataHelper class.
verified new applycal applymode='calflagstrict' and interp='linear,linearflag' options in the pipeline
CASA 4.2.2 pipeline release puch list finalized with PLWG
Swithched back to Flag Data
investigated pipeline plotms problems
worked with Brian Kent on pipeline flagdata tbuff support
Spent some time to understand claims that mosaic imaging is non reproducible
Image differences found to be systematically 1e-7 or better
understood, explained on the Jira why the results claimed happened and close the ticket
Completed the table based version of MSUVBin for large image..need testing
Fixed a couple of bugs and FTMachine unecessary image change in toRecord; updated SIImageMapper et al to save virtual model to disk rather than to record
Justo González Villalba
Implemented chan.avg/regridding section of new WEIGHT/SIGMA convention in mstransform and successfully cross-validated it vs cvel/split
Implemented time.avg section of new WEIGHT/SIGMA convention in AveragingTIV. I'm now in the process of cross-validating it vs split.
Sending out use cases for CARTA viewer
Several days out sick
Moving forward on both hiring positions
Worked on CARTA UI prototype.
Worked on pipeline plotms issues.
More work responding to tester concerns re: spxfit.
Created small test that shows Array<T>::operator*=() works as expected after receiving a report on casa-staff that it does not.
Responded to Urvashi's request re: input on proposed additional MFrequency::Type enum.
Requested tightvnc be installed on my system so I could login to OSX box.
Spent a lot of time investigating a performance issue with ia.imageconcat() reported for version 4.2.1.In the end, I could not reliably reproduce the issue on trunk, and so I could not address it.
Punted to Mark ticket about making immath param verification error messages more descriptive because this verification occurs at the tasking level before immath code is run.
Addressed an issue raised by tester re: impv.
Fixed labeling in imfit output.
Initial examination of some EVLA ionosphere test data taken June 7
Prep and delivery of "Visibility Weights in CASA" talk at CTF
Continued discussions with Justo re averaging in mstransform
Lots of JIRA triage
Working on FITS related issues
Added support for stokes imaging (other than I) and input shapes for Clark vs ClarkStokes minor cycles.
Added controls for running imager in 'readonly' mode or to save a model in the last major cycle ( virtual or modelcolumn )
Slowly cleaning up the JIRA backlog.
Ger van Diepen
A new cascara branch on googlecode has been made containing most NRAO changes. I'm working on merging in the latest changes, which seems to be an endlessly moving target until NRAO finally moves to the googlecode repository. The most important changes are:
All compiler warnings (including the newest g++ and clang) have been removed. Three important warnings are:
A for and if should always use braces. The compiler complains about ambiguous else if used nested without braces.
Some people don't know that in a logical expression AND has higher precedence than OR. Therefore the compilers warn for an if containing AND and OR without parentheses.
The compiler warns about unused named parameters. Thus don't give a parameter a name if not used.
Casacore is fully thread-safe (apart from a counted reference in MeasFrame). It means that the static variables/functions and (most) counted references are thread-safe. Similar to STL writing to the same object in multiple threads is not safe.
AIPS_CXX11 is set if a C++11 compiler is used.
CountedPtr uses std::shared_ptr (from C++11 or tr1), thus is thread-safe.
TaQL got many new features, most importantly GROUPBY/HAVING with aggregate functions.
The RO versions of Table/Scalar/ArrayColumn have disappeared (which was already the case for NRAO's casacore) and replaced by a typedef/#define. In the future the same will be done for the RO classes in MeasurementSets.
Worked on setjy modifications (more on refactoring of some of per-spw processing)
Updated a setjy unit test which was not properly testing currently available mode
User support - looked at an error in fluxscale reported by a VLA user. The exception occurred at the statistical calculation as only a single gain solution exists for one of spws. Modified the code to allow to proceed in such a case.
Friday NAOJ Meeting
Update issue type in single dish JIRA tickets.
Created three more tickets of CASA 4.3 targets.
Investigated an issue that sdimaging creates blank image for flagged data ( CAS-6346, related issue in synthesis imaging CAS-2961)
Discussion of printing warning for old SD tasks in CASA 4.2.2 ( CAS-6718)
Started a new monthly meeting with TP commissioning leads (Cat, Anthony, Tsuyoshi) at JAO to exchange latest information in each side. Takeshi, Bunyo, Hiroko, and George are also attending from NAOJ ( minutes)
Added a unit test for sdsave to check if fillweight parameter works correctly (CAS-6181)
Found that TIME values in the output of gencal look like starting time of integration, not midpoint as defined, and reported it to George and Tak.
sdstat: modified reference data set to replace by mathematical result using a numerical expression of summation. unittest is fine, checked in.
maintenance of PIPELINE automated verification shell script