ListPerm doesn't check its arguments, can give nonsense

[wilfwilson]

ListPerm is a global function that doesn't validate its arguments.

When given input that is valid (according to its documentation), it behaves fine. Unfortunately people don't always give valid input, such as me 10 minutes ago, when I forgot the brackets in a permutation and wrote ListPerm(5,20).

gap> ListPerm(5,20);
[ 1, 32, 243, 1024, 3125, 7776, 16807, 32768, 59049, 100000, 161051, 248832,
  371293, 537824, 759375, 1048576, 1419857, 1889568, 2476099, 3200000 ]
gap> last = List([1..20],i->i^5);
true
gap> last2 = OnTuples([1 .. 20], 5);
true

gap> ListPerm(1,10);
[ 1 .. 10 ]
gap> ListPerm( One(GL(3,5)), 20 );
[ 1 .. 20 ]

I suggest ListPerm should give an error if its first argument isn't a perm and its optional second argument isn't a positive (or non-negative?) integer .

[fingolfin]

I agree that these checks should be made.

In addition, perhaps we should also change OnPoints, OnPairs, OnTuples and OnSets to validate their input more. Because that's is the underlying cause for what you observe; essentially, OnTuples(T, g) is equivalent to List(T, x -> x^g):

gap> OnTuples([1..20], 5);
[ 1, 32, 243, 1024, 3125, 7776, 16807, 32768, 59049, 100000, 161051, 248832, 371293, 537824, 759375, 1048576,
  1419857, 1889568, 2476099, 3200000 ]
gap> OnSets([1..20], 5);
[ 1, 32, 243, 1024, 3125, 7776, 16807, 32768, 59049, 100000, 161051, 248832, 371293, 537824, 759375, 1048576,
  1419857, 1889568, 2476099, 3200000 ]

These kernel functions could reject the case where g is not either a perm / transformation / partial perm, or else an external object.

On the other hand, no serious harm is done by the current behavior?

[frankluebeck]

These function can be performance critical in some code. Before adding tests one should check thoroughly that the performance loss is negligible for all kinds of valid arguments.

[fingolfin]

If you mean OnSets etc: The test would be a simple TNUM test in the kernel in the code for non-permutations. I strongly doubt that would have a measurable overhead.

For ListPerm: if performance critical code relies on that, it probably should be rewritten...

[wilfwilson]

perhaps we should also change OnPoints, OnPairs, OnTuples and OnSets to validate their input more.
...
On the other hand, no serious harm is done by the current behavior?

I don't see any harm in the current behaviour of OnTuples etc, and moreover this behaviour is meaningful, and therefore potentially useful. Perhaps the documentation for these functions should be generalised to point out that they also work outside of the context of group actions.

[fingolfin]

I am fine either way, but I do think that ListPerm should raise an error if the arguments have nonsense type.