CASA User Testing Meeting Minutes - 2013-04-23, 09:00 Socorro / 11:00 Charlottesville

Agenda

Participants

Present: Mark Rawlings (Chair), Crystal Brogan, Adam Leroy, Jen Meyer, Scott Schnee, Jeff Kern

Apologies: Kim Scott, Bryan Butler

Locations

* Socorro: AOC-317; IP: 146.88.16.8

* Charlottesville: Room 331; IP: 193.33.119.19

Minutes (Main Points Only)

Second (pre-)release package (r23729) has been available for testing

casapy-prerelease 4.1.0 :23729

To launch it, type: casapy -r4.1.0

Test reports were due on Friday, April 19th.

A third pre-release package build is anticipated in the next couple of days or so. Mark will circulate an annonucement when it's available.

Launching the pre-release in Socorro

On Red Hat 6 computers in Socorro, users should type:

casa-prerelease

or failing that,

casapy-prerelease

Planning Cycle for the Next Release

Reminder: Meetings have taken place for ALMA, and VLA. The existing document has been updated. A Google spreadsheet containing details of CASA 4.2 priorities has been drafted.

Future User Tester Participation in These Meetings

It was felt that in the future, except during the run-up weeks to a new public release, user testers would only be expected to attend these CASA User Testing meetings if they were currently testing an item, and felt it needed a particular contribution. This proposed operating model is based upon the assumption that the agendas for each meeting will be posted on the preceding Monday, and that testers will review the agenda and update all their tracking tickets as needed in advance of the meeting. The only parties expected to routinely attend all the meetings would be Mark, Jeff, Crystal, Bryan and Juergen.

Of course, any testers still genuinely interested in attending would be most welcome...

Any Other Business

There was a generally held perception that it had become more easy to lose contact with the plotms window, and several flavors of failures were mentioned (these are documented in existing tickets, and there is already a fix committed for at least some of them).

The release notes need a few additions: see (e.g.) CAS-4616 (below). Also, warning notes regarding OS X/Qt issues should perhaps be placed on the downloads page. Additional such items will be discussed directly by the CSSC.

Current testing items/issues (see also Google spreadsheet workbook)

Spreadsheet available here

Mark will check with Juergen on the current status of the PV diagrams-related tickets. Notes will be placed here or in the ticket entry lines, as appropriate.

Open Blockers

  • CAS-4602: plotms: iteration through spw hangs in OSX 10.7 - Currently 'Open' (4.1 Release will not be held back for this: Mac OS X Qt issue)
  • CAS-4310: plotms freezes on OSX 10.7 while iterating on axes - Currently 'Open' (4.1 Release will not be held back for this: Mac OS X Qt issue)

Ready To Close

  • CAS-3857: FlagData / FlagCmd on Cal Tables ("Add support for the new CalTable format at the task level") - 'ARTT'. Most recent check-in to trunk was r23406 - Confirmed to be in in first pre-release package. Had been dependent on CAS-3565 (resolved). Confirmed during meeting: will be marked as resolved (at least from the CASA perspective).
    • Obs_Id and time-based scanning.
    • Crystal had looked at this for ALMA.
    • John Lightfoot has also looked at this.
    • CAS-5010, created following the previous meeting to specifically track Juergen's backwards-compatibility testing. Crystal had looked at this for ALMA and was happy with it. Awaiting a comment from Juergen, but will be closed.
  • CAS-4978: imfit box doesn't work as advertised - Currently 'ARTT'. Assigned back to Remy for testing. Bug fix (r23599) already in second pre-release package. Tested, and works as advertised, and answers to the remaining questions have been posted. Confirmed during meeting: will be marked as resolved.
  • CAS-4616: Narrow-band A-projection - Currently 'ARTT'. VLA item. Frazer Owen (plus?) had been looking at this. Lots of notes from the developers in this ticket. Probably at the final state it will be in for the 4.1 release. Discussion during meeting: ticket can be marked resolved, but an Action Item was placed on Juergen to provide advice in the release notes as to when this feature should be used.
  • CAS-4939: Vectorize calwt in applycal - (Previously asserted to be ARTT without checking by Jeff). Crystal had looked into this one. She was happy with it, but noted that calwt needs improvements in documentation.

New Approved for Testing Targets

  • CAS-4898: Testing the new vi/vb framework with virtual model columns - Currently 'ARTT'. This is the first item to use the new underlying vi/vb framework. Rather than assigning specific testers to it, Jeff just asked test users to watch out for and report any strange or unexpected behavior when working with the virtual model columns. It would be very helpful if several testers could run flagdata( vis = 'xxx.ms', mode='clip', clipminmax=[0.1,10.0], datacolumn='residual' , mode='calculate' ) on any MS on which they have previously run 'clean' with usescratch=False. This command will not change any flags in the MS, but it will test the use-case we are concerned about. Since nothing is being written to the MS, a seg-fault should not corrupt anything in the MS itself.

Ready for Retesting

(N.B. Items marked ALERT! are awaiting the third pre-release for re-checking).

  • CAS-4964: impv fails complaining of stokes axis issue - Currently 'ARTT'. To be tested by Juergen, with Adam as a watcher. Included in current pre-release package. Adam thought that this now looked OK. Pending the nod from Juergen, I suspect that this one can be closed.
  • CAS-4921: refactor image statistics code - Currently 'ARTT', with Mark Rawlings assigned. Confirmed during meeting: will be marked as resolved as soon as it's had a final re-check in the third pre-release.
  • CAS-1893: Cal Tables in PlotMS - Currently 'ARTT' again. Ongoing. Latest updates from Patrick & George (r23341) were merged into pre-release, so this was fixed again in the first pre-release package. Ticket reassigned to Jeff to look into the current state of affairs with this one.
  • ALERT! CAS-4632: UVContSub: Ability to specify line corrupted channels - Currently 'ARTT', with Scott Schnee assigned. The bulk of the testing on this is done, and this is just awaiting a final check-back on the latest version of the documentation in the third pre-release before closing.
  • ALERT! CAS-1880: Multipanel plotting - Currently 'ARTT'. Lots of feedback given. The developer reports that he's working on addressing issues found. Some fixes put in. Had difficulty reproducing the partial page example. Awaiting third pre-release.
  • ALERT! CAS-5011: Source finding in viewer - 'Currently 'ARTT'. Testing done by Adam. Some fixes put in, judgement to be made based on third pre-release.
  • ALERT! CAS-4829: Create method which returns smallest beam all beams in an image can be convolved to - Currently 'ARTT'
    • When the dictionary is used, the parameters 'major', 'minor' and 'pa' should not also be set (which by default, they currently are. Fix put in, awaiting third pre-release.
  • ALERT! CAS-4930: Add new method to msmd tool: intentsforfield() - Currently 'ARTT', Todd had been testing this, and given feedback. Slowness issue (see below) being addressed in another ticket (fixes put in for third pre-release). nUnflRows inconsistency?
    • The current version of listobs does not report Off Source (subscan) intents. This is in contrast to msmd. listobs should report the number of unflagged rows (and this feature is covered by the now-resolved CAS-4197). It does, but has been very slow (because it has to run through all of the dataset) - see also CAS-5038. Because of this, the need for a flag option to switch off the counting of unflagged data is covered in CAS-4943 (see above). CAS-4930 is also relevant to this topic. See also CAS-5047 (below).
  • ALERT! CAS-5047: taql queries have the possibility of making msmd run very slow, so reimplement such queries as iterator loops. Currently 'RTT' - needs tester assigning. Fix committed, expected to be availabile in third pre-release.
  • ALERT! CAS-3383: Addition of combine=ObsId in Solver - Currently 'ARTT'. Crystal was testing this for ALMA, and was happy with it (modulo a review of the documentation). Latest updates to this will be in third pre-release.
  • ALERT! CAS-4881: Investigate impv issues discovered during testing. Reset to 'ARTT'. Juergen has provided feedback. Fix put in, recent docs fix committed which should be in the third pre-release.
  • ALERT! CAS-5049: Viewer sundries parent ticket. Currently has 19(!) subtasks. Deemed to be a very important issue for testing of the third pre-release.
    • CAS-4962: imregrid: Provide some examples of use cases for imregrid for documentation purposes - Currently 'Open'. Some examples provided, and used as basis for new (already drafted) documentation. Can probably be closed soon.
  • ALERT! CAS-5034: for 4.1, add warnings that imregrid ignores shape and axes if regridding to new reference frame - Currently 'RTT'. Assigned to Mark for testing with third pre-release.
  • ALERT! CAS-5057: imregrid "get"/template mode broken in r23729 - Currently 'RTT'. Fix committed, awaiting third pre-release.
  • ALERT! CSV-2709: sdlist fails for high spectral resolution single dish data - Effectively 'RTT', and awaiting third pre-release.
  • ALERT! CAS-5060: The use of cvel is making uvcontsub (actually its clean) run incorrectly - Previously flagged as 'Resolved' by developer, but needs checking. Blocker issue. Fix should be in both trunk and the third pre-release.
  • ALERT! CAS-5051: peel needs to be removed - Currently 'RTT'. Assigned to Juergen. Blocker issue. Awaiting third pre-release.
  • ALERT! CAS-4927: imhead reimplemented mode="list" ready to test - Currently 'ARTT', with Mark Rawlings assigned. Testing done, fix committed. Awaiting third pre-release to just recheck.

Items Currently Being Tested

  • CAS-4754: Create a new framework with a new task and tool to replace applycal concat, cvel, partition, hanningsmooth and split - Currently 'RTT'. Parent ticket with many subtasks that are marked as 'Resolved'. This parent ticket details a number of MSTransform features that can be exercised, and documentation's provided. Ticket refers to availability in release branch 23427, so should be in the current pre-release.
    • NOTE: This is currently being tested by someone in Socorro, but will be hidden/not be advertised as available for the 4.1 public release. It will be tested more seriously after the 4.1 release.
    • From previous meeting: Juergen was to look into general ia tools testing for PV diagrams (e.g. "Does smooth fail gracefully?", etc.)
  • CSV-2594/ CAS-4970: CASA plotms fails with a segmentation fault when averaging channels - Currently RTT. Testing to be done by EA staff in Chile. Fix recently committed to trunk (r2805).
  • CAS-4953: Testing of Mac OS X 10.7 build under 10.8. Currently 'ARTT'. Jen's been testing this extensively. Remy Indebetouw's been playing with it as well. Jen may want to comment. Parent ticket for a couple of subtasks. Some plotms-related issues reported.
    • CAS-5065: imview under Mac OS X 10.8. Currently 'ARTT'. Lost of feedback given, and responses from the developers.
  • CAS-4705: Implement "feather" in simalma - Currently 'ARTT'. Kim Scott has provided a test report. Some questions raised about image fidelity of output?
  • CAS-4614: Feathering GUI Improvement - Currently 'ARTT'. Juergen & Ed Fomalont have been testing this. Ongoing.
    • Ed had previously felt that it was not yet ready for general use. Science was to provide a punch-list of remaining issues to be fixed (in code/documentation/both).
  • CAS-4297: reimplement imhead - Currently 'In progress'. Currently parent ticket for CAS-4921 and CAS-4927.

  • CAS-4943: add parameter to listobs/ms.summary() to turn off counting of unflagged data - Currently 'ARTT', but see also next section. No tester formally assigned?

Upcoming Test Items

...

Developer Addressing Test Results

  • PV Diagrams
    • CAS-4599: Interactive creation of PV-Diagrams in the Viewer - Currently 'ARTT', with assignment to Juergen, who has provided some feedback. It's the parent ticket of CAS-2222 (itself a duplicate of CAS-1032). Juergen (and perhaps others) may want to comment on this one. Can hopefully be closed fairly soon. Darrell's working on it now.
    • CAS-2996: Task interface to perform arbitrary P-V slices through data - Currently 'ARTT'. There's been quite a bit of back-and forth on this ticket between Juergen and the developer. Question for Juergen: is this now complete for the 4.1 release? Can hopefully be closed fairly soon.
  • CAS-5059: pclean failures in 4.1. Currently 'Open'. Kumar is looking into this. Action Item on Jeff to talk to him about its current status.

Input Required

  • CAS-4715: Image masked / zeroed when regridding from GALACTIC to J2000. Question for Science: What to do about padding image prior to rotation?
  • CAS-2852: Tasks that produces a histogram of values in a region of an image and fits a gaussian to the histogram - Currently 'Awaiting Input'. This one's still awaiting input from Jeff Kern.

Other issues

  • CAS-4671: This is partly due to the way setjy currently process multiple spectral windows (i.e. set the model for each spw one by one). If clearing of the model at the beginning of the setjy is introduced, it would solve the issue but would limit functionality of other cases. The problem is in the current underlying C++ method (can perhaps be fixed, but doubtful for 4.1).
  • imregrid
    • CAS-4959: imregrid: When using image templates, the default shape used should be that of the template (not the input image) - Currently 'Open'. To be addressed via improved documentation for 4.1.
      • CAS-4960: imregrid: When regridding, the regridding should by default be done in accordance only with whichever axes are common to both the template and the image (and are sensible) - Currently 'Open'. To be addressed for 4.1 with improved documentation only.
      • CAS-4961: imregrid: axis ordering of data cubes appears to be inconsistent - Currently 'Open'. Much recent discussion as to what can be included for 4.1. To be addressed for 4.1 with improved documentation only.

Running List of Action Items

Standard Template Format:

  • Item Description:
    • Assignee:
    • Date Assigned:
    • Due Date:


-- MarkRawlings - 2013-04-18
Topic revision: r6 - 2013-04-23, MarkRawlings
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback