Re: how does print-to-codes work?

[Jasper Taylor]

On Thursday 12 Sep 2002 11:17 am, Daniel Diaz wrote:
> Try to use get_print_stream/1 inside your portray predicate: like:
>
> portray(F) :-
>         float(F),
>         format_to_codes(Fs, "~6f", [F]),
>         append(Ns, Os, Fs),
>         \+ (member(Ch, Os), \+ member(Ch, "0.")), !,
>       get_print_stream(Stream),
>       format(Stream,"~s",[Ns]).
>
>
> This should work

Yes, it does, thanks very much -- my mistake for not reading the manual 
thoroughly enough. 

However I am getting some bizarre behaviour -- the behaviour of print_to_codes 
seems erratic, and this causes problems with my application. Here is an 
example I tried on two different machines (both RedHat 7.3) to show you what 
I mean, and note that varying the order of calls even a little can change the 
outcome...nicefloat contains the predicate listed above...

$ gprolog
GNU Prolog 1.2.14
By Daniel Diaz
Copyright (C) 1999-2002 Daniel Diaz
| ?- print(2.92).
2.9199999999999999

yes
| ?- print_to_codes(A,2.92).

A = [50,46,57,49,57,57,57,57,57,57,57,57,57,57,57,57,57,57]

yes
| ?- consult(nicefloat).
compiling /win98/Program Files/Simile/Develop/Tests/nicefloat.pl for byte 
code...
/win98/Program Files/Simile/Develop/Tests/nicefloat.pl compiled, 9 lines read 
- 1616 bytes written, 8 ms

yes
| ?- portray(2.92).
2.92

yes
| ?- print(2.92).
2.92

yes
| ?- print_to_codes(A,2.92).

A = [50,46,57,50]

yes
| ?- print_to_chars(A,2.92).

A = ['2','.','9','2']

yes
| ?- print_to_codes(2.92).
uncaught exception: 
error(existence_error(procedure,print_to_codes/1),top_level/0)
| ?- print_to_codes(A,2.92).

A = [50,46,57,50]

yes
| ?- print_to_chars(A,2.92).

A = [à,á,'\x12\','B',à,á,'\x12\','B']

yes
| ?- print_to_codes(A,2.92).

A = [224,225,18,66,224,225,18,66]

yes
| ?- print(2.92).
2.92

yes
| ?- print_to_codes(A,2.92).

A = [224,225,18,66,224,225,18,66]

yes
| ?-


Any workaround would be welcome !!
        --Jasper