Check for scalar cubes before attempting to concatenate a CubeList

[valeriupredoi]

📰 Custom Issue

Hi guys, a variation on the theme of concatenation/merge on which there are a few issues already open - it'd be nice if iris handled scalar cubes as big bad wolves and looked for them in, say, a CubeList to be concatenated/merged, and raise a nice exception that identifies that (or those) scalar cubes - I recently found out that scalar cubes produce havoc when one tries to concatenate - raised (inadvertently) by @bouweandela in this comment - it all happens in _io._concatenate_overlapping_cubes(cubes) - after extract_time() (one of our utilities) the resulting cube may be scalar - so iris concatenation fails due to scalar cubes not having dim_metadata attributes in iris._concatenate_CubeSignature class (they do, but they are empty lists).

Cheers muchos 🍺

[pp-mo]

#3234 seems relevant

So, we also labelled "Feature: merge+concat"
It's not clear when we can do this,
But let's leave it open for now.

ALSO: can consider this + similar issues around time of forthcoming ESMValTool workshop [Oct 18-20 2022]

[valeriupredoi]

cheers @pp-mo 🍺 Indeed, let's add it to the workshop!

[valeriupredoi]

@pp-mo I have created a discussion point on the ESMValTool discussions pages that can hold a list of all these types of issues to be discussed at the call on next Monday, and the wkshp, if you got a bit of time it'd be nice to populate that list with what you reckon could be discussed - or, label these issues with some specific label so I can pull them over there. Cheers muchly as ever 🍺 ESMValGroup/ESMValTool#2792

[pp-mo]

... label these issues with some specific label so I can pull them over there.

We do have the "Feature: ESMValTool" label.

I just added that one to a few more existing Iris tickets.
I think that now covers everything we know about so far, that is of relevance to ESMValTool i.e. one of you expressed interest.
Feel free to add to other issues if they have interest for you, and we didn't already know it.

[pp-mo]

We do have the "Feature: ESMValTool" label.

@bjlittle also reminds me -- we also have a Scitools project, though TBH we haven't paid much attention to it lately.

[bouweandela]

We discussed this issue a bit in the ESMValTool community, and the request is mostly about a nicer error message. @valeriupredoi Could you provide an example of the error message and explain why it is not clear and how you think it could be improved?