We attended parts of the Pipeline Working group meeting in Garching last week. Based on several discussions with Dirk M. and Stewart, we plan to do:
run parallel tests on tclean using the mfs mode, which should work without convergence issues;
verify how to run the pipeline in parallel when invoked with pipelineMakeRequest and then the PPR (use-case from Felix);
verify a failure (probably a bug) that Stewart found when trying to run sessions in parallel that called flagdata. The issue is that flagdata is internally parallelised, but the sessions approach would try to fork jobs in MPIServers, but currently it can only do this from an MPIClient. His use-case for these sessions come from the SD pipeline, which will have many ASDMs to be processed all together.
Working on CAS-10142: add support in flagdata to take the exposure time into account when flagging a timerange. This led me to using the git framework in a different way. I need to use a branch from a fork in casacore. Should I add this information to the Plone git page?
Discussions about Dave's new implementation of statwt2 and the integration into mstransform.
Continued work on overplotting atmospheric transmission curve CAS-9053
Visit to Socorro. Working and discussing about a number of topics:
Extending the VI/VB2 to cope with TVIs that need to access several SPWs simultanously.
AsyncIO support in VI/VB2 and its impact on mstransform.