Re: Linux 9.0.1: boot -all *almost* works

[email protected] (Tim Bunce) Fri, 28 Dec 2001 02:52:13 +0000
Newsgroups perl.dbi.oracle-oci
Message-ID <[email protected]>
On Thu, Dec 27, 2001 at 05:58:43PM -0600, Stephen Clouse wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> There is difficulty grokking this (from ociap.h):
> 
> sword OCIExtractSetKey(dvoid *hndl, OCIError *err, CONST OraText *name,   
>                        ub1 type, ub4 flag, CONST dvoid *defval,
>                        CONST sb4 *intrange, CONST OraText *CONST *strlist);
> 
> This is the XS generated for it:
> 
> sword_status
> OCIExtractSetKey(hndl, err, name, type, flag, defval, intrange, strlist)
>         void *  hndl   
>         OCIError *      err
>         OraText *       name
>         ub1     type
>         ub4     flag
>         void *  defval
>         sb4     &intrange
>         OraText * const &strlist
>         OUTPUT:
>         intrange
>         strlist
> 
> Predictably, this gets thrown:
> 
> Error: 'OraText * const' not in typemap in OCI.xs, line 14552
> 
> But fix the typemap, and:
> 
> OCI.c: In function `XS_Oracle__OCI_OCIExtractSetKey':
> OCI.c:16601: warning: initialization from incompatible pointer type
> OCI.c:16601: parse error before `const'
> OCI.c:16604: warning: passing arg 8 of `OCIExtractSetKey' from incompatible 
> pointer type

Ah well, unless you _need_ OCIExtractSetKey just add it to the skip
list in the boot script.

Thanks for the report. I'd guess that the const handling code in h2xs
needs tweaking.

FYI, I'm fairly close to a DBI release and if I'm _really_ lucky
DBD::Oracle and Oracle::OCI might just follow soonish afterwards
as I'm trying to get slightly more up to date before Perl Whirl '02.

Tim.