Mb/explicit

[sesquideus]

Finished converting the InputSets and Input classes to the implicit detection of tag parameters (detector, target, ...).
Also added more test to cover for new edge cases (not many though).

[chyan26]

I guess there are issues remains with the simulation data. Since there is no conflict with current code. I would suggest we rebase and merge it first.

[sesquideus]

Yes, also on my other machine everything works and passes all the tests now... but I do not understand the actual problem.

Still, I think that the benefits of merging now (even without these four tests passing) outweigh the problems it will cause if we diverge too much.

[hugobuddel]

I would prefer that we at least identify the problem before merging anything. Could someone else reproduce the problem?

[sesquideus]

My bad.
I did not realize I fixed my pyesorex ("if not" → "is None"). The recipes that are failing are exactly those without parameters.

[hugobuddel]

I've made a bug report, https://jira.eso.org/browse/PIPE-12044 , and two PR's, one with some tests https://gitlab.eso.org/cpl/pyesorex/-/merge_requests/18 , and one with your fix, https://gitlab.eso.org/cpl/pyesorex/-/merge_requests/19 . I don't think you can access those, but I thought it would be good to add a link to them anyway. I can put the branches also on our fork.

But it shouldn't be too hard to work around this bug for now right? Maybe we can automatically add a dummy parameter to every recipe or something like that?

I much prefer adding such an ugly workaround than having broken tests.

[hugobuddel]

Or perhaps we can keep these lines in?

    parameters = cpl.ui.ParameterList([])

I also want to note that I still had problems even with your fixed pyesorex. Because wouldn't removing this line result in parameters to be None? As they are set to None in the base class IIRC. So the is None would still be trigger, just like the if not was, right?

[sesquideus]

Yes, we should keep this line in, as a system-wide MetisRecipe default.

If I understand it correctly, cpl.ui.ParameterList([]) is Falsy, and thus it crashes within pyesorex. But it is not None and that works for me now.

[hugobuddel]

By the way, in my bug report I noted that the first recipe on https://www.eso.org/sci/software/pycpl/pycpl-site/user/basics.html does not specify any parameters, so it actually is None there, so that case should be fixed as well (but that is not our problem).

I added a suggestion to add a dummy parameter; would that work for now?

[hugobuddel]

I pushed my test branch to our fork as well, see https://github.com/AstarVienna/pyesorex/tree/addtests .

(By the way, I deleted all the nonsense branches in the pyesorex fork, but they keep turning up again, so need to figure out what is happening.)

[sesquideus]

Here we go 🥇

[hugobuddel]

It would be good to also add these new recipes to the esorex and EDPS tests, but we can do that later.