problem with strictly sorted inhomogeneous lists

[ThomasBreuer]

The following happens in the current master branch as well as in some released versions of GAP.

gap> l:= [ 1, Z(2) ];;
gap> SetIsSSortedList( l, true );
Error, filter not possible for list (plain,dense,nhom)
not in any function at *stdin*:10
type 'quit;' to quit to outer loop
brk>

This is strange if one looks at the following.

gap> 1 < Z(2);
true
gap> l:= [ 1, Z(2) ];;
gap> HasIsSSortedList( l );
false
gap> IsSSortedList( l );
true
gap> HasIsSSortedList( l );
true

The second piece of code shows that GAP regards the list as strictly sorted,
and also stores the filter.

Here is a situation --in the master branch-- where one runs into trouble because of the error.
Note that Filtered assumes that its result is strictly sorted whenever the argument is strictly sorted, and wants to apply this knowledge.

gap> l:= [ 1, Z(2) ];;
gap> IsSSortedList( l );
true
gap> Filtered( l, IsObject );
Error, filter not possible for list (plain,dense,nhom) in
  SetIsSSortedList( res, true ); at .../gapmaster/gap/lib/coll.gi:1380 called from 
<function "Filtered">( <arguments> )
 called from read-eval loop at *stdin*:15
type 'quit;' to quit to outer loop
brk> 

[fingolfin]

This issue is already present in GAP 4.4.12, so I'd argue it was intentional. Of course we are free to revise this. But it's definitely not a recent regression.

[ChrisJefferson]

I think (I'm guessing pre-history) there was a time when "Sets" were supposed to be homogeneous, but now we don't require that any more. Certainly I think #3401 fixes it, and doesn't cause (I think) other problems.

[ThomasBreuer]

I do not understand @fingolfin 's comment.
It cannot be intentional that on the one hand, GAP can find out with IsSSortedList that a given list is sorted, and then store the filter, but on the other hand, GAP refused to set the filter via SetIsSSortedList.
Either inhomogeneous lists should never get the filter HasIsSSortedList, or SetIsSSortedList should be admissible for inhomogeneous lists.