Re: scm_integer_p and scm_num2int mismatch

[Dirk Herrmann]

On 7 Oct 2001, Jan Nieuwenhuizen wrote:

> The behaviour of scm_num2int has changed, it now only wants an integer
> value as input, but the behaviour of scm_integer_p has not changed, so
> this no longer can serve as a test for valdid scm_num2int values.

We've had a discussion about the issue some days ago.  I don't claim that
there has been a definitive solution about it, but to me it seems that
sort of a consensus was:

* scm_num2int and friends are going to be deprecated.  The reason is that
  the names are irritating:  They suggest that any number should be OK as
  an input parameter, although handling non-integer values requires some
  transformations.  It was decided against performing such transformations
  implicitly.  (Not sure about this one:  The old behaviour of
  scm_num2int is going to be restored during the deprecation phase.)
* instead, functions like scm_integer2int and scm_exact_integer2int will
  be provided.  The first one would also accept inexact integers and
  implicitly coerce them to exact ones.

Best regards
Dirk Herrmann