Re: Additions to the gh_ interface.

[Dirk Herrmann <dirk@ida.ing.tu-bs.de>]

On 26 Sep 1998, Jim Blandy wrote:

> > However, there is also need for a test, whether a Scheme number is
> > convertible to a C builtin number type. I assume that such a test will
> > only be required in combination with the actual conversion. Therefore, I
> > suggest the following solution: 
> > 
> > long gh_scm2long(SCM obj);                /* the current function */
> > 
> > long gh_safe_scm2long(SCM obj, int* flags);   /* the new function */
> > 
> > The status of the conversion could be returned in the flags variable.
> > (Instead of int, an enumeration could be used, or a typedef, ...)
> 
> Hmm... that's an idea.  The conversion process does produce the
> information that you want to put into *FLAGS, as a side effect.
> Granted that Guile should provide some way to discover whether a
> Scheme number can be accurately converted to a particular C type, is
> this the best interface?  What would you do if you got an OVERFLOW,
> besides raise an error?

Well, you could 
- issue a more detailed error message with SCM_ASSERT
- do some cleanup on the heap before actually raising the error
- just provide a default value to perform the computation with

You could even decide to perform the desired computation with guile's
bignums. ...hm...think... Hey, that raises a good question: are guile's
bignums accessible from the C level? If so, then there could also be a
conversion gh_scm2bignum or similar. I think some days ago someone
suggested that guile used some GNU bignum package. In this case, the user
could provide two implementations: a fast one with C data types and a
fallback with bignums.

Best regards, 
Dirk Herrmann