Tom Lane wrote:
> KaiGai Kohei <kaigai@kaigai.gr.jp> writes:
>> However, I wonder if the oid field should be also preserved at same
>> place, not inside a specific trigger function.
>
> Possibly. I wasn't planning to mess with it now; but if you've fixed
> the other problems with assigning to a system column then maybe we
> should allow it for OIDs too.
I moved the code to preserve system attributes in my tree, as follows:
http://code.google.com/p/sepgsql/source/detail?r=1190
The patch set does not allow to update "oid" column *now*, so the condition
of above if-block is always true. But the writable-system-attribute feature
can be implemented in similar way.
Thanks,
--
OSS Platform Development Division, NEC
KaiGai Kohei <kaigai@ak.jp.nec.com>