The problem is that the test case tests whether the code handles trimmed CCDs correctly, but it does not test that the crosstalk task is called in the correct place.   We have just redefined when crosstalk is corrected, so there are two consequences:

• The test case needs to be changed to handle an untrimmed CCD, with both parallel and serial pre- and post-scan pixels
• We need a test case that exercises crosstalk within the isrTask.

The latter might be done by using ci_hsc (but only after DM-15161 is resolved), or by using a new ci_lsstCam (DM-15163).  A stand-alone test for the isrTask is also doable, but involves faking a lot of things or providing quite a lot of real data.

I take this ticket to only cover modifying test_crosstalk.py to test the new definition of the crosstalk task; is that correct, John Swinbank?

As the ticket is written, I'm not sure that is correct. The ticket says

Please augment the test case to fail when things are run in the wrong order.

which I take as being a request for an integration test for the interface between assembly and crosstalk correction. However, I didn't intend to be dogmatic about that: I have no objection to splitting this ticket into two or more pieces along the lines you suggest (in that context, I note that I'm not sure if your first bullet is adequately addressed by the trivial changes made to the test case on DM-15152).

On the other hand, I do wonder if it wouldn't be adequate to address this by simply have the crosstalk correction code raise if it's fed a trimmed CCD. That would certainly be easy to test!

The DM-15152 changed caused the measureCrosstalk code to break, (DM-15410) and this wasn't detected because the tests were not augmented to cover non-trivial CCD geometry.