Comment author: administrator

Full_Name: Wheeler Ruml
Version: 3.06
OS: linux
Submission from: katsura.parc.xerox.com (13.2.18.21)

Scanf doesn't seem to like the floating-point format used by
string_of_float. string_of_float is happy to print nothing after the
decimal point, but as the following code demonstrates, scanf will
raise an exception if the string ends directly after the decimal
point:

let scanf_test () =
Printf.printf "got %f.\n" (Scanf.sscanf " 24. " " %f" (fun x -> x));
(* now no space after the point *)
Scanf.sscanf " 24." " %f" (fun x -> x)

scanf_test ();;

got 24.000000.
Exception: Scanf.Scan_failure "scanf: bad input at char number 4: bad int".

Thanks for the bug report: this behaviour was due to a wrong treatment
of the end of input condition and has been corrected some time ago
into the working sources.

I suppose one could argue that string_of_float is matched with
float_of_string and scanf is supposed to be matched with printf, and
the two pairs shouldn't necessarily be combined, but it seems like an
unnecessary incompatibility.

I don't clearly grasp what you mean, here. What is the incompatibility
you want to get rid of ?

Thanks for some great software!

Thank you again.

Best regards,

Pierre Weis

INRIA, Projet Cristal, Pierre.Weis@inria.fr, http://pauillac.inria.fr/~weis/

Comment author: administrator

Fixed by PW in 3.06+...