Thanks Paul I was I will apply the patch. The is the garbage collection
part of the code so I was expecting some little glitches like this on
different boxes. I will apply the patch. I guess we will have a RC5 next
week
> I've been playing around with the DBD::Oracle 1.20 RC4 embedded object
> support and found that I get segfaults unless I apply the following patch:
>
>
>
> --- oci8.c.old 2007-12-13 16:54:45.625233000 -0500
> +++ oci8.c 2007-12-13 16:48:21.204639000 -0500
> @@ -1544,7 +1544,7 @@
> fld = &obj->fields[pos]; /*get the field
> */
> if (fld->typecode == OCI_TYPECODE_OBJECT
> || fld->typecode == OCI_TYPECODE_VARRAY || fld->typecode ==
> OCI_TYPECODE_TABLE || fld->typecode == OCI_TYPECODE_NAMEDCOLLEC\
> TION){
> empty_oci_object(fld);
> - if (SvTYPE(fld->value) ==
> SVt_PVAV){
> + if (fld->value &&
> SvTYPE(fld->value) == SVt_PVAV){
> av_clear(fld->value);
> av_undef(fld->value);
> }
> @@ -1570,7 +1570,7 @@
> default:
> break;
> }
> - if (SvTYPE(fld->value) == SVt_PVAV){
> + if (fld->value && SvTYPE(fld->value) == SVt_PVAV){
> av_clear(obj->value);
> av_undef(obj->value);
> }
>
>
> I don't really know the code well enough to understand why fld->value is 0
> in those cases, but it was. I hope someone who knows this code will look
> at it.
>
> -Paul
>
