On Fri, Sep 05, 2008 at 09:46:57PM +0200, Jeffrey Ratcliffe wrote:
> http://www.sane-project.org/html/doc011.html#s4.2.6: Access to a
> scanner is provided through an opaque type called SANE_Handle. The C
> declaration of this type is given below.
>
> typedef void *SANE_Handle;
>
> While this type is declared to be a void pointer, an application must
> not attempt to interpret the value of a SANE_Handle. In particular,
> SANE does not require that a value of this type is a legal pointer
> value.
>
> So in the typemap I have:
>
> SANE_Handle T_OPAQUEPTR
>
> I assume that newSVpv(h, 0) is causing me the trouble. I can't work
> out what is correct.
newSVpv() is documented as:
Creates a new SV and copies a string into it. The reference count for the
SV is set to 1. If C<len> is zero, Perl will compute the length using
strlen(). For efficiency, consider using C<newSVpvn> instead.
SV* newSVpv(const char *const s, const STRLEN len)
So yes, it's doing a strlen() because you specify the length as 0. To do what
you wanted to do, you wanted newSVpvn():
Creates a new SV and copies a string into it. The reference count for the
SV is set to 1. Note that if C<len> is zero, Perl will create a zero length
string. You are responsible for ensuring that the source string is at least
C<len> bytes long. If the C<s> argument is NULL the new SV will be undefined.
SV* newSVpvn(const char *const s, const STRLEN len)
But I'm not even sure that that's sensible. Your typemap is
> void
> sane_open(class, name)
> SANE_String_Const name
> INIT:
> SANE_Status status;
> SANE_Handle h;
> PPCODE:
> status = sane_open(name, &h);
> XPUSHs(sv_2mortal(newSViv(status)));
> 126-> XPUSHs(sv_2mortal(newSVpv(h, 0)));
So what you're doing is a moderately expensive way of creating an empty string.
You're not storing the value of h itself.
Assuming that you wanted to store the value of h, I don't know what the best
thing to do is, and I have to stop typing now as food is served.
Nicholas Clark
