Socorro: Jeff, George, Kumar, Rob, Tak, Susan, Sanjay, Urvashi, Jim
CV: Darrell, Dave,
News / Meetings / Visitors
Ken Sowinski going to half time next month.
NAOJ Visit July 28 - Aug 1
Build, Release, Testing
JIRA update to 6.3.1
Pipeline (Lindsey) - 7/24.
Flagging (Sandra, Justo or Urvashi) in October? * VI/VB2 (Jim) - with a focus on integration effort - possibly August.
CASA 4.4 Development Targets Input
4.2.2 RPMs (and Tarball) available
Still testing AW-Projection, need to turn on rotation
Fighting against memory footprint
Found a way to correct w-term for large mosaic by making use of the correlator phasecenters..will be published in MosaicFT
made MSUVBin independent of synthesis which was the last piece remaining in dependence of mstransform on synthesis.
Justo González Villalba
Re-sync changes in msvis re-modularization with 4.2.2 merges
Advanced in implementation of OBJECT ref. Frame in mstransform
Prepared list of 4.4 targets for Rob
Debugged together with Sandra flagdata issue reported from ALMA (FLAG_ROW inconsistent with FLAG cube) . I've come to the point where I can proof that flagdata is ok but the data served by VI/VB already has the problem. I'm trying to determine now if the corruption happens at read time or at write time (in previous buffers) due to non-contiguous data access.
Located and fixed bugs associated with MSVis Independence.
OODT Deployment for VLA Pipeline
Viewer crash for complex images.
CARTA linked views.
Participated in discussion for summer students re: what one can do with a PhD in astronomy other than astronomy.
Spent a couple of days looking at memory management issues and imstat (CAS-6759). Concluded that python isn't garbage collecting memory of local variables like I assumed it was. Made a few fixes, but a leak remains because I don't understand why python is not reclaiming memory due to local variables in a task function after a task has returned.
Participated in new imstat filters requirements discussion. Began looking at associated code to start design phase. Spent a lot of time looking at how the image statistics framework currently works. It's actually very nonintuitive and complex and spans many casacore classes. With that information, I thought a lot about how to add new filters in a way that will support current ia.statistics() parameters, minimize modifications of existing code, minimize maintainence costs, and minimize impact on our casacore partners while still using the same general framework that is already in place.
Investigated CAS-5204 imfit chan range use user-provided estimates, and found that, contrary to reporter's belief, imfit does use the user provided estimates for each channel and does not use the results of the previous channel's fit as the estimate for the current channel. Resolved ticket as Won't Fix.
previous week (I was absent from last week's meeting)
Worked with Rob on critiquing new imstat requirements.
Fixed test_ia_convolve() and imagetest_regression broken tests.
Investigated CAS-6749 (imhead: mode='list' does not work for decimated images with 'restfreq). Was unable to reproduce reported issue.
Responded to Rob re: 4.4. priorities for image analysis
Added NCP projection support to imhead, CAS-6568. Awaiting Dirk's response about my suggestion of adding DirectionCoordinate::isNCP() rather than his suggestion of using a much more complicated existing static method in FITSCoordinateUtil that requires the caller to determine and pass in a lot of information.
Added support to return median area beam to imhead, CAS-6727.
Passed image cross/auto-correlation ticket to Rob for algorithm input.
Replied to casa-staff re: query about build issue regarding casacore/casa/version.h.
Put fix in test_imcollapse for table that was being left open after run.
Investigated and fixed test_rg_fromtextfile.py failure which was due to undefined spectral reference frame support added in r30403. Fix was to update test fixture used to check region data structure.
Because ia.statistics() was tapped as a possible reason for the test_sdimaging failure on casa-staff, I investigated but this test completed without errors for me on a clean build in my sandbox, so I could not reproduce the Jenkins failure.
Investigated odd flag state behavior at Sandra's request (CAS-6746, reports from several ALMA users), on suspicions it was a rendering problem in plotms. In fact, FLAG_ROW is somehow getting set (in flagdata executions) in rows for which all chans/corrs are not flagged, and it depends on the order in which flagdata runs are executed. Sandra currently suspects VI2.
Investigated lazy filler data shape error (submitted CAS-6766). Turns out ACs have correct shape (and are correctly expanded from compressed format in BDF!), and all XCs have wrong shape. Michel is on the case.
Studied finer points of VI2/VB2 deploy in calibration in the solve context, which requires OTF averaging and thus must use TransformingVI2 nesting and MSTransformManager. Looks like a mstransform dependence on synthesis is nominally implied. Discussed this with Jim and Kumar a bit.
Continued work on WEIGHT_SPECTRUM deploy
Investigated a gaincal crash reported by pipeline testers, but could not reproduce it (***Liz H. reports Jul 21 that she is now unable to reproduce it... Argh.)
Correspondence with Jeff and Jason Kooi re Jason's return visit in late August to put final touches on basic Ionosphere corrections for 4.3 release.
Some work on v4.4 dev priorities
Imager Refactor : Finished restoring beam features (auto-detecting good beam when requested) --- Added uv-taper option to weighting --- Enabled psf-patch calcs in Clark clean --- Added option to restart clean from existing residual images and skipping first major cycle --- Added masks to spectral index images.
Fixed CAS-6776 about non-editable masks in interactive clean due to spectral frame mismatches. This bug keeps appearing and this time it was declared loudly by an irate user at my door as a 'fatal flaw in casa' :).
Made EL6 builds and relinked to be usable on non casa machines.
Made request list for 4.4.
Working with Dave and others to settle on the requirements for new noise estimating algorithms in imstat.
Compiling team input for 4.4 development targets.
Admin fun: CARTA contract, ADMIT contract, QSU exception reports, etc.
Initial check-in for 'setpol' functionality in setjy after some testing and fixing some bugs (using 3C391 data set from the Casa Guides). ToDos: write unit test(s) and add the description of the added parameters in the help doc.
Desktop replacement - rebuilt casa and also took this opportunity to clean up and back up the disk contents
Friday NAOJ Meeting
Shinnosuke Kawakami after removing async, remaining unittest are almost passed