Broken IsTrivial for empty semigroups

[fingolfin]

Without semigroups loaded:

gap> M0:=Magma(FamilyObj([1]), []);
<commutative semigroup with 0 generators>
gap> IsTrivial(M0);
false
gap> Size(M0);
0

But with semigroups loaded, I get an error due to the last line of this method provided by semigroups:

InstallMethod(IsTrivial, "for a semigroup with generators",
[IsSemigroup and HasGeneratorsOfSemigroup],
function(S)
  local gens;

  if not IsFinite(S) then
    return false;
  fi;
  gens := GeneratorsOfSemigroup(S);
  return IsIdempotent(gens[1]) and ForAll(gens, x -> gens[1] = x);
end);

I guess a simple fix would be to add Length(gens) > 0 and to the condition in the return. And I assume a test case should be added, too?

[james-d-mitchell]

@fingolfin thanks for the report, the empty semigroup rears its head again! Happy with your proposed fix, also will be happy to implement this if you don't have time

[fingolfin]

I would appreciate if you could take care of it. I am never sure which branche(s) I should target with fixes. Also, not sure where you'd like to have a test case.

BTW, we noticed this because it broke a new test case in GAP master.

[olexandr-konovalov]

The broken test can be seen on travis here: https://travis-ci.org/gap-system/gap-docker-master-testsuite/builds/342310829 (that's a travis cron build which now performs testinstall/standard/bugfix with three combinations of packages: none, default and all loaded.

[james-d-mitchell]

Fixing it now, and, of course, happy to do so.

For the record @fingolfin I think we are using the same branch structure as GAP itself, bug fixes go into stable-3.0 (at present) and new features into master, and for every file in the semigroups/gap directory and its subdirectories there is a test file in tst/standard/ which contains tests for that file.

[james-d-mitchell]

This is resolved by PR #460.

[fingolfin]

Thanks @james-d-mitchell . But for the record, this is not how GAP works: We do all work on the master branch, and then cherry-pick selected fixes to the stable-4.9 branch (for a few weeks, we did it differently, but that was mostly an exception for a few bigger changes we still wanted to get into GAP 4.9).

Also, you have two (semi-)active stable branches, 2.8 and 3.0 (or at least you had -- but perhaps now 2.8 is closed), which added to my uncertainty -- i.e. did not know into which subset of 2.8, 3.0, master you want this fix. So I thought it'd save everybody time to just mention the issue and leave it to you guys to know where you want this :-).

Anyway, I'll submit PRs again in the future, too, and if I select the wrong branch, I am sure you'll guys will let me know and I can then simply rebase :-).