Using (SCM)0

Using (SCM)0

[Ludovic Courtès]

Hi,

Is there anything wrong with using (SCM)0 as a "special value" in C
code?  In other words, is it possible for a "real" SCM object to be
equal to zero, in which case I'd better use SCM_UNDEFINED or some such?

Thanks,
Ludovic.


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: Using (SCM)0

[Marius Vollmer]

Ludovic Courtès <ludovic.courtes@laas.fr> writes:

> Is there anything wrong with using (SCM)0 as a "special value" in C
> code?  In other words, is it possible for a "real" SCM object to be
> equal to zero, in which case I'd better use SCM_UNDEFINED or some such?

Yes.  (SCM)0 will be interpreted as a pointer to a cell at address 0,
which will segfault.

You need to be careful with SCM_UNDEFINED as well.  I must _never_
reach Scheme code.  I.e., you can't put it into pairs, vectors, etc.

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: Using (SCM)0

[Ludovic Courtès]

On Fri, Oct 10, 2003 at 04:38:56PM +0200, Marius Vollmer wrote:
> Yes.  (SCM)0 will be interpreted as a pointer to a cell at address 0,
> which will segfault.

Sure.  Actually I meant the following: Suppose I have a structure with
an SCM field and I use calloc () to allocate it; can I write in my code
something like:

  if (object->field == (SCM)0)
    /* Unitialized, assume there is no "real" Scheme object here */
  else
    /* Assume object->field is a "real" Scheme object */

Thanks,
Ludovic.


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: Using (SCM)0

[Marius Vollmer]

Ludovic Courtès <ludovic.courtes@laas.fr> writes:

> Sure.  Actually I meant the following: Suppose I have a structure with
> an SCM field and I use calloc () to allocate it; can I write in my code
> something like:
>
>   if (object->field == (SCM)0)
>     /* Unitialized, assume there is no "real" Scheme object here */
>   else
>     /* Assume object->field is a "real" Scheme object */

Yes, you can do this (tho it might be easier to initialze the field
with SCM_BOOL_F once, say).

However (and I didn't think about this before) the type SCM is not
necessarily an integer or pointer.  It should be treated as opaque.
You need to use SCM_PACK:

  if (object->field == SCM_PACK (0))

but there is no real guarantee that SCM_PACK(0) is identical to
all-bits-zero.  Hmm.  It seems cleaner to avoid testing a SCM for
zero.

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user