Use collection-2021-1.0 conda env for tests

[mrakitin]

Here is the update to run against the collection-2021-1.0 conda environment.

Expected outcomes:

• failed test with the latest jedi v0.18
• after collection-2021-1.0 NSLS-II/profile-collection-ci#16 is updated to disable jedi (i.e., with the --IPCompleter.use_jedi=False option, see https://github.com/NSLS-II/playbooks/pull/411), we should see the successful test.

[mrakitin]

Actually, it's not that trivial to test the tab-completion programmatically. @jklynch, any idea about that?

The failed test was due to another reason.

[jklynch]

Maybe we have to be satisfied with triggering the error some way other than tab completion, and trying to recognize the specific exception.

[jklynch]

It makes no sense to me, but it really looks like there is a colon missing from the PV that is causing the current failure. Should there be a : after {Scr:1}?

[mrakitin]

It makes no sense to me, but it really looks like there is a colon missing from the PV that is causing the current failure. Should there be a : after {Scr:1}?

That worked fine with collection-2020-2.0rc9: https://dev.azure.com/nsls2/profile_collections/_build/results?buildId=1260&view=logs&j=f9902ac2-07f5-5f38-a83c-04fda0b3ca53&t=f8d34d36-0f4a-596c-c42f-6c04499af201&l=782. I suspect it's something broken in the latest version ophyd, or we just misconfigure the timeouts.

[jklynch]

We had to downgrade pyepics from 3.4.0 to 3.3.0 for a similar problem I think. That may have been two deployments ago, so it might not be relevant any more.

[jklynch]

Is it possible to add a caget XF:12ID1-BI{Scr:1}cam1:Acquire_RBV to the test?

[jklynch]

Is the spoof-beamline failure relevant?
https://dev.azure.com/nsls2/profile_collections/_build/results?buildId=1262&view=logs&j=c58b40dd-1ba0-5a8c-8e7a-987c20afdefa&t=8f1b40f4-c2fc-5e37-c197-a69f5d504b9f&l=736

[mrakitin]

Is the spoof-beamline failure relevant?
https://dev.azure.com/nsls2/profile_collections/_build/results?buildId=1262&view=logs&j=c58b40dd-1ba0-5a8c-8e7a-987c20afdefa&t=8f1b40f4-c2fc-5e37-c197-a69f5d504b9f&l=736

Oh, good eye! Looks like it can cause trouble. I wonder why the responses were successful in prior files, before 26-...

Is it possible to add a caget XF:12ID1-BI{Scr:1}cam1:Acquire_RBV to the test?

Yes, I specifically added this line some time ago to support such drop-in requests/tests:
https://github.com/NSLS-II/profile-collection-ci/blob/9a729913d22289322f3ddca72ff6aa695f4c8bf6/azure-linux.yml#L195

[mrakitin]

Well, that drop-in.py file will not be executed, as it will fail earlier. I'll add another .sh file to support it.

[mrakitin]

This error may be a sign of deeper issues. Today at SRX I discovered the non-existent PVs were still a part of the class, which did not fail in the previous versions of ophyd for some reason, but failed with the currently deployed version 1.6.0 (see NSLS-II-SRX/profile_collection#24). I guess the class may need the updated cam, image, roi* components with updated subcomponents for AD_V33+.