blfit does not work with CABB data

[JamieStevens]

During the last reconfigure, data was taken simultaneously with both the ATCA correlator and CABB. Using blfit on the ATCA correlator data gives sensible results, but the CABB data produces nonsense.

So far I have identified the following symptoms (or causes, I don't know yet):

• the errors on the solutions are much too big, allowing blfit to assume that the solution is that all antennas are exactly where they are supposed to be
• the seeing monitor values (smon internally) are always the same value, and these values do not correspond to the values present in the ATCA correlator data

I am trying to follow and debug the code by running the ATCA data and CABB data through gdb and finding differences that are suspicious, but have yet to hit upon anything definitive.

[JamieStevens]

After a bit more testing, it appears that excluding auto-correlation data (using select=-auto prior to inputting to blfit) makes blfit work properly on CABB data.

It may be beneficial to alter blfit so that it ignores auto-correlation data if it is present.

[VincentMcIntyre]

See also ticket:82

[MarkWieringa]

Since then, the noauto option in atlod has been restored, making it easy to avoid loading the autocorrelations. This is probably sufficient.