PlugAlgo : Add array[1] -> scalar conversion

[johnhaddon]

This conversion is available in the various query nodes, and is particularly useful when querying attributes and primitive variables from certain DCCs which like to write everything as arrays, even if they make more sense as scalars.

I've only implemented this for FloatPlug, IntPlug and BoolPlug so far. If there's general support for the idea then I'll update to include other types too. Part of me dislikes the conversion and wishes everything was just typed more logically in the first place. And I worry that the existence of the conversion only for length 1 might lead to a surprise or two for things with animated array lengths. But on balance I think this is probably the most pragmatic thing to do - it lets users deal easily with a bunch of common problem scenarios without resorting to the performance hit of expressions.

[danieldresser-ie]

Yeah, I think I'm in full agreement, both with finding this somewhat uncomfortable, and also in agreeing that this is probably the most pragmatic approach.

The code LGTM.

[murraystevenson]

Ugh. This is quite unfortunate but I'd go for it as it will make things easier for users...

And I worry that the existence of the conversion only for length 1 might lead to a surprise or two for things with animated array lengths

I wonder if returning the first value, rather than no value for lengths > 1 would be more or less surprising in that situation?

[tomc-cinesite]

I wonder if returning the first value, rather than no value for lengths > 1 would be more or less surprising in that situation?

+1 for 'the first' if you pick a scalar output for an array input.

[johnhaddon]

I wonder if returning the first value, rather than no value for lengths > 1 would be more or less surprising in that situation?

I'm far less comfortable with applying this to lengths > 1. In my mind we're supporting cases where the user intended to provide a scalar but it has been incorrectly encoded in an array. If there's more than one element in the array then that clearly is not the case, and it's far more ambiguous what the intention might be.

[tomc-cinesite]

I'm far less comfortable with applying this to lengths > 1. In my mind we're supporting cases where the user intended to provide a scalar but it has been incorrectly encoded in an array. If there's more than one element in the array then that clearly is not the case, and it's far more ambiguous what the intention might be.

Fair point.

[johnhaddon]

I've updated the PR with the equivalent conversions for all the other plug types now.

[murraystevenson]

LGTM 👍

[johnhaddon]

Rebased to fix merge conflict, and merging...