Socorro: Jeff, Lindsey, Pam, Martin, George, Tak, Kumar, Urvashi
CV: Ville, Darrel, Jen, Anand, Andy Akeem, Bob, Dave,
Garching: Sandra, Erique, Fredrico
News / Meetings / Visitors
No meeting next Monday, February 20 -- US holiday
Jeff has accepted the role of SRDP director. Formally, his last day as CASA lead will be March 20. In reality there will be a transition period beyond March 20 in which he is assisting his successor with getting up to speed.
OSX1012 and OSX1011 intermittent ic2233 errors.
RHEL6 test server host down
RHEL7 test issues resolved
It appears we will have to release CASA 4.7.2:
In pipeline there is a bug that affects an indeterminate number of projects in a bad way.
Configurations for ALMA cycle 5 were not uploaded properly into the repository.
4 Mac Minis and a new Linux server for build/testing uses have been cleared to order
Jeff has reviewed and approved Andy's task test list. I'll begin putting in JIRA tickets for developers to create task level tests that don't currently exist
A couple of questions for some existing tasks: history/are they very useful?
plotprofilemap - makes spectral profile map from specified image; the task accepts both CASA image and FITS cube as an input.
Tak was the last person to work on this.
find - Find string in tasks, task names, parameter names.
Sandra was the last person to work on this.
Anand will bring these tasks up for discussion at the next stakeholders meeting.
tclean test timeout issue - still under investigation
Do developers run regression tests from inside CASA?
The script that does this is currently not working in the new framework - is there any need to update it?
Dave does run these tests. (publish summary)
Darrell will add Instructions for doing this to Plone.
Future work - getting the CASAguides running in bamboo
Akeem will discuss with Darrell. This requires import of data into repository.
Setjy is assigned to Tak. If something new is needed, we will talk w/ Tak.
Tclean - task page is as ready as it will be for a while.
Description of items to be added in future will be left in document.
Urvashi will prune task links on the synthesis imaging page.
CASA Doc command: doc command is in. DS does not have a mapping from task names to URLs.
Issues with log origin getting mixed up in pipeline logs when running in parallel (CAS-9548). Some tasks do not set log origin, which might be the reason for the mix.
Accepted tests in Bamboo (HPC-related tests):
alma-m100 serial version with clean fails in Bamboo due to changes in peak and RMS of phase calibrator images.
alma-m100 parallel version with clean passes. TBV the differences between this version and the one above.
alma-m100 parallel version with tclean fails in SynthesisImagerVi2.cc, line 324): "Error in building Coordinate System and Image Shape".
ALMA pipeline testing:
Latest parallel run on full pipeline failed in stage 19 with the same error as above (CASA and pipeline v39532). SynthesisImagerVi2.cc, line 324): "Error in building Coordinate System and Image Shape".
EVLA pipeline testing:
The long VLASS parallel run got stuck due to the setting of detailed flag summaries (CAS-9330). Brian suggested I should set to False the parameter in hifv_applycals(flagdetailedsum=True). Restarted the run with this change, but it stopped in the same place. TBV.
The following staff need to take action this week:
You can also view your CASA Docs tasks in JIRA using the My Plone Tasks filter.
ALMA observing session support
EVLA polarization calibration task / heuristics
VLASS image list generation heuristics
ALMA outlier flagging heuristics
Matplotlib 'agg' backend support
VI/VB2 issues, setting export VI1=1 seems to bypass these
JIRA ticket management
Updating ESO Jenkins tests to use new casa-toolset.
CASA matplotlib backend 'agg' support discussions
CASA 5.0 testing, version checking, casa configuration and startup, matplotlib batch plotting
Improved ALMA reference antenna determination heuristics
CAS-9482 Ready to Verify (plotting two datacolumns in plotms)
CAS-9650 Ready to Verify (bug fixes when plotting two y-axes in plotms)
CAS-8050 bug fix for channel selection with averaging in ms tool, after input from Enrique
with layered VIVB2, need to set channel selection in ChannelAverageTVI (spwExpr in configuration Record) and set FrequencySelectionUsingChannels in VI2
Plone docs on browsetable, plotants
Created certificate for gdb on OSX
CAS-9501 Selected region in plotms shown in wrong place on OSX 10.11 and RH7 --> must be Qwt 6.1 problem
Reviewing ChannelAverageTVI. While looking for a fix to CAS-9441, it seems that there is a need to review the configuration options for ChannelAverageTVI ( CAS-9679).
A quick look shows that several classes are affected: FlagDataHandler, FreqAxisTVI, ms_cmpt, ConvolutionTVI, FreqAxisTVI, PhaseShiftingTVI, RegriddingTVI, UVContSubTVI.
Validation of UVContSubTVI with the new denoising lib. Review of the the problematic cases found by Jennifer ( CAS-8347). No differences found so far with uvcontsub or uvcontsub3. More detailed tests needed.
Investigated test_importasdm failure on RH 7 bamboo tests (appears to have been a resource problem on the test machine)
Continued to dig in to the flagger code (bdflags2MS).
Some work on the ASDM move out of the ALMA code tree.
Jeff has written a response to the user committee. Comments on this are due today. Chair is pushing for use cases that he wants. Excited about CARTA.
Federico M Pouzols
Working on CAS-5808 (timedev and timedevscale options for rflag), and CAS-9548 (logs issue in parallel/serial)
Looking at recent bug reports around mstransform: CAS-9695 (slicer error with combinespws), and CAS-9693 (slow split with spw selection).
explored artifact repositories
explored options of moving data repository from subversion
cleaned up CASA startup messages
Fixed a bug in the new pruning code for automask. Checked in.