Crash Reporting System Requirements Circulated (Martin and Julian to follow up).
Warnings (see summary page) improve by a huge drop in the synthesis module. Other modules held steady except MSTransform which increased by one.
Tests everyone write at least on c++ test
HPC
Development
Pipeline crashing on large cubes and Pipeline execution time
Pipeline
AOB
Developer Reports
Monday Meeting
Sanjay Bhatnagar
Sandra Castro
Lindsey Davis
Pam Ford
Plotms export work in progress. Can successfully export a screen-resolution plot some of the time, but learned interesting things about plotms threads and plotter factory.
Provide feedback for recently opened split issues, in principle these are issues reported from 4.4 which have already been fixed for 4.5 (multiple channel selection).
Improved mstransform task documentation for re-gridding section
Working on a new feature for mpi4casa to dynamically change at run time OMP settings.
Jim Jacobs
Jeff Kern
Architecture Tool Evaluation
Usual Meetings
Susan Loveland
CAS-8068: File permission problems with viewer
CARTA: Stack management, layer color maps, bug fixing, selenium tests
David Mehringer
Worked out with CIS a few glitches after upgrade to new hardware.
(CAS-8048) listobs requires 6 minutes to list an ALMA dataset with 12 minutes worth of 4ms dumps: WIP. Submitted casacore github pull request after improving performance of MSSummary on Monday, still awaiting review and merge.
Fixed test_imregrid failure on OSX after it was reinstated to automated test suite.
At Urvashi's request, axed logging messages from ia.pixelvalue() which unintentionally started appearing after a refactor I did a couple of weeks ago.
(CAS-7980) uid://A001/X121/X1bf weblog crash in final makeimages: RTT. Specified max amount of contiguous memory required by the underlying lattice iterator. Previously, for some PagedImages, contiguous memory equal to the size of the entire pixel array was being required.
(CAS-8073) image has trouble with second axis: RTT. Added a message to imfit in cases where convergence occurs but fit is nonsensical.
Discussion about pixel to world coordinate conversion issues in some cases with GBT images.
(CAS-8072) CRTF file parser does not correctly parse restfreq information in some cases. Resolved. Fixed two issues in which certain global lines without comma delimiters were being tacitly and incorrectly processed and cases where, on global lines some spectral specific parameters were specified but range was not specified.
(CAS-7986) Create MS Consistency Validation Tool: RTT. Added optional check parameter to ms.open() which is false by default. listobs now opens the MS with check=true. Checking code is in msvis/MSVis/MSChecker since Ger objected to these checks being put in casacore.
George Moellenbrock
CAS-8063: Diagnosed peculiar emergent behavior that occurs in calibration when concat-ing datasets with different/moved antennas and different spws.
Some discussion with Masaya about completing visstat2
Progress on using MPI in tclean; C++ build OK both with and without MPI; Python layer needs some work yet.
[VLA: Dealt with more disk usage issues at site.]
Urvashi Rao
added/modified python level tests for tclean
CAS-6937 : added trap for condition reported in ticket (change shape and restart tclean with same image name), and tried out few ways of handling 'overwrite' of images on demand.
Darrell Schiebel
Using origin as runtime loader seems to work.
Ville Suoranta
Julian Taylor
started looking at remaining differences between cvel and cvel2 (CAS-7259, overlapping channels flagged)
reported upcoming gcc 6 crashing on casa (PR68145), fixed now
Takahiro Tsutsumi
Continued investigating mask generated from CRTF file for clean/tclean (CAS-7121). Found that the issue was in regionmanager. Created a new ticket, CAS-8072 and Dave fixed it.
Fixed a bug for the case when spw with channel selections, start= and nchan is not default (=-1) (CAS-8077)
Cleaned up some of the compiler warnings from GBT related FITS code under code/nrao. Then realized that these are probably not in use. There are some more single dish codes (for the140ft and old 12m?) in nrao module which are the sources of the warnings. Maybe we can remove them from CMakefile at least.