[std-interval] More on interval computations as proofs

Guillaume Melquiond guillaume.melquiond at ens-lyon.fr
Sat Sep 30 20:37:37 PDT 2006


Le vendredi 29 septembre 2006 à 22:21 +0200, Gabriel Dos Reis a écrit :
> | An alternative implementation is to have a global flag like errno,
> 
> I would not support that.  Global variables pose significant problems,
> both for users and implementers -- these days errno really is a
> function.  Specially in multi-threaded environment (C++ is seriously
> considering concurency).
> 
> Another alternative design could throw an exception.

This would make the library a bit painful to use. Indeed, when the flag
is set, it only means that the computations do not satisfy the
hypotheses of "fixpoint" theorems. The computed range, however, is still
valid and the program may have a use for it. So, if an exception is
thrown, the only way to obtain the range will be to start the whole
computation anew with no-exception functions.

Best regards,

Guillaume



More information about the Std-interval mailing list