On usage of commonIdea

[JacquesCarette]

commonIdea and commonIdeaDict should check if the list they are given is empty and, if so, complain. mkIdea should be used instead.

There are quite a few abuses of this in Drasil, and they should all be fixed.

[hrzhuang]

Is there a reason why AuxConsProg uses a CI instead of a type class constraint like NamedIdea?

[JacquesCarette]

Probably no good reason for it. Try to fix it and see what happens!

[hrzhuang]

So the problem here is that AuxConsProg constructs a AuxConstntSec, which is used in DocSection. If we use a type class like NamedIdea for the first argument of AuxConsProg, then AuxConstntSec would need a type parameter, which in turn would require DocSection to have a type parameter, and this seems to be a change that would require some more deliberation.

The CI argument to AuxConsProg seems to just be used for the name of the program (e.g. GlassBR). Does it in fact make sense to use a raw CI here?

On a related note, a number of uses of commonIdea/commonIdeaWithDict with empty domain is just the names of the programs. If it is appropriate for these to be CIs, what should be their domains?

[hrzhuang]

Another "abuse" of commonIdea is the smart constructors dcc' and dccWDS' for CommonConcept. I suppose if CIs should always have a non-empty list of domains, then that list of domains should also be added as an argument to dcc' and dccWDS'?

[JacquesCarette]

Right - things like DocSection should have much more concrete contents. Having it have an extra type parameter seems like it could spiral out of control (though it wouldn't be wrong).

Your follow-up question is excellent: we need a name, not a 'concept' here. The only reason we'd want a concept is to be able to cross-reference the UID somehow. Which might happen!

As to: what is the 'domain' of a program name? This, I don't know. Which maybe is a further hint that this isn't a "concept" per se.

There is a bigger issue with these representations - I can't find the number right now. @balacij ?

[balacij]

Sorry, I'm not entirely sure what the larger issue you're referring to is, @JacquesCarette. Is it about the "gradual chunk building?"

[JacquesCarette]

Yes, and the project to revisit how we store the data in chunks.

[balacij]

Gotcha, that's #2873. We also have a larger discussion in chapter 8 of my master's thesis.