Comment (by Roland Haas):

Looks mostly fine with me. Please apply. An actual pull request would have been nice though since it would have made it easier to inspect the code and verify the code in context.

For example the lower bound of the loop is somewhat off since the loop starts from 1 so the first val_next inspected is gr->group(1) which has already been assigned to val befor in val = meval(gr->group(1),eedata). This does not lead to incorrect results (since min(a,a) = a) but is a bit odd.

The test change is good to apply as well because, while it does remove a test for assigning a simple float to a parameter, there is another such assignment in

TestPar::real1[2] = 42.

ideally the tests would also include a call of min(42.) ie a single argument to min or max.

--
Ticket URL: https://bitbucket.org/einsteintoolkit/tickets/issues/2257/piraha-return-in-wrong-place