Bad IsDualSemigroupRep implication

[fingolfin]

This code is problematic:

InstallTrueMethod(IsDualSemigroupRep,
                  IsSemigroup and IsDualSemigroupElementCollection);

A representation is type information / encodes the internal layout of an object; it is not a mathematical property and as such should not be implied in this way.

This is in a way a return of the problem from issue #556 and is the main blocker for the GAP PR gap-system/gap#3006

[james-d-mitchell]

Thanks for the report @fingolfin but I'm fairly sure we've fixed this already in:

a194195

Despite the misleading name, IsDualSemigroupRep is a category not a representation, and will be renamed in the next major release of Semigroups.

Maybe I need to make a release to incorporate the change in the commit mentioned above, I can do that!

[fingolfin]

Ah, sorry, the above was based on a CI run from a few weeks (?) ago, so perhaps there was already a newer release that contains the fix.

Sorry for not checking more carefully!

[james-d-mitchell]

No problem, I'm not sure I've released a version with that change yet, but I can do it this week.

[fingolfin]

There is no need to hurry. I plan to release GAP 4.12.1 this week, but I don't think there is a need to include this new release of semigroups; if it is "too late" it'll just be in 4.12.2 which I right now hope to release in mid-December

[james-d-mitchell]

Resolved in 5.1.0