Socorro: Jeff, Lindsay, Jim, Pam, Susan, Martin, Juergen, Kumar, Tak
CV: Anand, Ville, Darrell, Akeem, Andy, David, Jen
News / Meetings / Visitors
May 17, ANASAC (Socorro)
May 18-19 NRAO Users Committee
May 23-27 PLWG Face 2 face (No MMM)
May 30: NRAO Holiday (No MMM)
June 1-8: Synthesis Imaging Summer School
On Friday, Josh enabled proper permissions for the new bamboo installation. Both the new JIRA and bamboo servers are running. subgit has been working okay.
RHEL7 tests are running now. Current errors agree with RHEL5 & 6 tests.
Something that Kumar committed broke simulation. Kumar is working on it.
M100 Analysis Regression failure on OSX 10.11
Cleared up once data was updated, thanks Sandra!
MPI long test run time. Andy will call Julian to resolve this.
Slowdown in imaging regression test run time
Kumar saw this on 2 tests: wprojection_regression and orionmos4sim_regression. Kumar will refresh his build and see if problem persists.
Andy only sees a slight slowdown on wprojection_regression
His initial suspicion was work Sanjay was doing on vi2/vb2 interface, which did not pan out
Re-assigning some tests from Jeff to other developers
Andy has requested some investigation by Jim.
Jen is waiting on Jeff for direction on how to handle image analysis documentation JIRA tickets. A technical problem has been encountered that is preventing David's recent updates from being available in Python.
Jeff recommends closing tickets since the documentation has been updated.
Delivery will happen in conjunction with the documentation update project.
The CASA Warnings Summary shows flagging back to green, with modules alma, singledish and synthesis still with warnings.
CAS-8595 Resolved (baseline iteration and locate on cal tables causes plotms crash)
CAS-8484 RTT (plotms high-resolution pdf export)
CAS-8408 RTT (plotms flagged points default now red circles instead of none)
Investigated CAS-7817 (averaging while iterating over spw) - created CAS-8612 for inconsistent weight/sigma spectrum values from VIVB2 depending on spw selection, weight/sigma okay
Beginning work on CAS-8270 (implement plotcal in plotms)
Jeff has requested a write up on sync function.
Justo González Villalba
With Ville's help we've gotten the crash reporter building and capturing the crash information and assembling a tar file for transmission to a server. This week I'll see if I can get the server folks to get that side online so we can get it running end-to-end.
Investigated cas-6119 (memory leak in memory tool). Discoverd that our implementation of Swig bindings leaks like a sieve. The leak is small, but in a loop making lots of tool calls to methods having string, variant or record valued parameters will eventually run out of memory. The JIRA reporter was calling measures once per visibility in their script. I've found and fixed it to first order, but need to complete a regression run to make sure that something didn't get busted in the process.
Continued pushing on the Atlassian Software
CAS-8591: Viewer shows wrong value on top axis when velocity is around 0.
CARTA: Multiple color map issues fixed.
(CAS-8426) im.pbcor() should write history to image it creates: RTT. Example in docs updated as requested.
(CAS-8341) Write history to new image when running ia.fromrecord(). IR. It was requested I add to the docs a reason why this method might be useful. I don't know any such reason since I inherited ownership of this method and there is no design nor requirements documentation as to what the motivation was for its creation.
Method was intended for inter-process communication in parallelization. It will be used again when MPI is active.
User has requested an example of use. Jeff recommends stating only that this is a serialization-deserialization method.
(CAS-8608) ia.setmiscinfo() should write to image history. RTT.
(CAS-8607) ia.setcoordsys() should write to image history. RTT.
(CAS-8605) ia.setbrightnessunit() should write to history. RTT
(CAS-8397) ia.calcmask() to write history. Resolved. Explained why extra "\" escapes appeared in python display of returned list.
(CAS-3847) Global Image module object to record memory allocation/de-allocation. IR. Assigned to Sanjay so that, upon his return, he could respond to Ger's follow up questions.
Fixed imagetest failure caused by modification of ImageMoments code.
CAS-8533: Verified ALMA polarization calibration in CASA 4.6
Continued calibration test program work
VLA Tours for local 4th grade
Tests of effect of tile shape on IO performance. Completed on nmpost cluster with Lustre filesystem, continuing testing with local disk.
Julian is looking at effect of tile shape on overall pipeline.
tweaks to pipeline's runsetup2/setup.py for Brian Kent's testing
completed scripts for generating Understand C++ database
RHEL7 tests are setup and may be working (nightly builds have been working for some time)
Got crash reporting working on Mac
First Carta package for OSX
Started working on Carta packaging on Linux (builds and runs on RH7, crashes on exit on RH6).
Met on Friday with Josh and Darrell and got the proper access and permissions for the new Atlassian suite.
Spent the most time on CAS-8584, the "tclean NaN issue". With help from Kumar, we found that tclean's hogbom code was missing image padding in deconvolution. Kumar made modifications for the padding and also added check to detect divergence, which made divergent behavior go away for one of the test data sets. But It did not completely fix the issue as the tclean run with another data set that Lindsey provided went into an infinite loop. Still seems to be missing a trap for some diverging cases.
Fixed the bug that automask was not executed when tclean is in interactive mode
Friday NAOJ Meeting
investigating slowness of sdimaging ( CAS-8562): test using new MS-based procedure revealed aother issue, the slowness of applycal. applycal takes several hours to apply calibrations to one fast-scanning MS (~700MB) which only takes 10min by ASAP. Coud it be related to tiling issue CAS-8415?
fitmode='auto' (line detection by line finder) in tsdfit ( CAS-7906): behavior of line finder verified.