j6626
commented
7 years ago Multiple SPWs are useful for continuum though....
I'm in favor of vis_sample only accepting a single measurement set. I think that's perfectly reasonable. And for line data that's been regridded and with a consistent number of channels, vis_sample ostensibly should be able to handle multiple execution blocks as is. And I think it's ok to oblige the user to only enter measurement sets where the entire dataset is supposed to be interpolated (e.g. no extraneous data just hanging out in another SPW).
CASA has a bug where the same source can be assigned different field numbers for different execution blocks. So forcing the user to select a single spw/field combination is suboptimal. I think it's best just to stick with warning messages in case the dataset of interest is pathological like that.
Should either create parsing and loops for handling these cases directly, force users to select a single spw/field combination, or at least display a message to the user that vis_sample can only handle a single spw/field combo at a time.
I'm in favor of restricting to a single spw and displaying an error message. There are no use cases I can think of where multiple spws are useful for line data, other than multiple execution blocks, and in that case concatvis could be used prior to vis_sample I think.