fix indentation when printing double cosets

[ThomasBreuer]

The distribution of \< and \> hints was obviously wrong.

I still do not understand why the hints were chosen this way. Usually the idea is to mark positions in the string where line breaks are more suitable than in other places, for example at the commas that separate objects; but here it is the other way round?

resolves #5840

[fingolfin]

This is indeed weird. I would expect it to start with \> not \<. More like so

Suggested change

	return(STRINGIFY("DoubleCoset(\<",
	return(STRINGIFY("DoubleCoset(\>",
	ViewString(LeftActingGroup(d)),"\<,\>",
	ViewString(Representative(d)),"\<,\>",

[ThomasBreuer]

The point is that I do not know what the idea behind this code was.
Also the fact that whitespace behind the commas is missing is unusual.
I guess this code was never really used, otherwise the indentation problem would have been noticed earlier.
Shall we just change this method?

[fingolfin]

Shall we just change this method?

Fine by me.

[fingolfin]

This obviously is better than what was before, but still looks strange (similar for the ViewString method for (right) cosets)

[ThomasBreuer]

Another irritating behaviour:
When debugging the problem, I tried the following:

gap> G:= SymmetricGroup(5);;
gap> H:= SymmetricGroup(3);;
gap> K:= SymmetricGroup(2);;
gap> dc:= DoubleCosets(G, H, K);;
gap> meth:= ApplicableMethod( ViewObj, [ dc ], 1, "all" );;
gap> meth[1]( dc );  # a method from "teaching mode" ...
"TRY_NEXT_METHOD"
gap> meth[2]( dc );
Error, SetPrintObjIndex: bad state, PrintObjDepth is 0 in
  SET_PRINT_OBJ_INDEX( i ); at /usr/local/lib/gap-rsync/lib/list.gi:3886 called from 
<function "ViewObj for a finite list">( <arguments> )
 called from read-eval loop at *stdin*:7
[ 

Calling the ViewObj method explicitly runs into an error, whereas the same method works when it gets called from View?

[fingolfin]

The kernel operations ViewObj and PrintObj involve some special magic in order to handle printing of recursive structures, i.e. of lists and records which directly or indirectly contain themselves. So that e.g. l:=[];l[1]:=l; gets printed as [ ~ ].

The printing methods implemented in GAP code call the kernel function SET_PRINT_OBJ_INDEX for this purpose, but that is only allowed "during" a ViewObj resp. PrintObj invocation.

That's why the method does not work when called in isolation.

[fingolfin]

This fixes an error so I'll merge it. Perhaps the printing can be improved further but that can be done in a follow-up PR

[ThomasBreuer]

O.k.

I had a look at the \< and \> handling in general. There are not too many places where these hints occur, but they are not consistent.

(Also, documentation is missing, and comments in the code are talking about "indentation", which I find confusing.)