New Hire Qian Gong - HPC Software Engineer, beginning in the Fall
MSv3 Report Closeout
CASA 5.6 - release branch today
Can't run parallel tclean from wheels in current form
Propose: after 5.6 branch and switch to 5.7 on trunk, modify tclean to use ia.imageconcat if custom concat binary not available
Use modified tclean in 6.0 release (small deviation from 5.6)
Upcoming stakeholder meetings : July 31 (summary of 5.6, 6.0, CNGI), Aug 7 (5.7 planning), Aug 14 (5.7 implementation discussion).
5.6 work after the branch
CAS-12433 and CAS-12590
5.7 plans underway.
New features : Focus on items needed critically for VLASS operations, ALMA operations, and the ALMA correlator upgrade. When possible pick minimal effort route, as we still anticipate moving largely to CNGI from next Fall.
For other non-critical dev items, focus on specifications in preparation for CNGI development
More verification tests.
Support Casa 6 migration and (inevitable) debugging.
Calibration and Flagging
casa 6 use of refimageconcat ?
Import / Export
New test machines having performance issues. Servers are currently down, including casa-metrics
Need to figure out what to do with the verification data sets and tests. Lustre+cvpost65?
General concern: CASA 6 is not thread safe. casatools/logsink and all casa tasks (and anything else using a common casatools/typecheck/validator) are not thread safe. See last comment of CAS-12552 for an explanation and possible fixes.
CAS-12552 needs to be merged to master as soon as possible!
The gRPC binding to casaplotms is generally complete. However, a number of plotms tests still fail. The previous binding for DBus was poor and did not take advantage of the capabilities provided by the selected tools. As a result, the new binding differs significantly from the original. In addition, the gRPC always uses a server thread unlike DBus.
Assembling the casa6 all-inclusive distro will begin this week. This includes mixing in the other misc GUIs (logger, table browser, etc.) and creating a combined distribution.
Build / Testing
5.6 validation list:
CAS-12433 tclean with chunks has discrepant PB, restoringbeam, and spectrum at edges of subimages (includes CAS-12432 but NOT CAS-9852), can be merged after branching
ONLY NEED DOCUMENTATION OR VERIFICATION TEST UPDATES:
CAS-12148 Serious issue with explicit setting of restoringbeam parameter in tclean
CAS-9108 Linear basis-specific 'Xf' (multi-scan) improvements