Re: Wrap access to Oid in HeapTupleHeader

On Mon, 01 Jul 2002 09:46:41 -0400, Tom Lane <tgl@sss.pgh.pa.us>
wrote:
>Manfred Koizar <mkoi-pg@aon.at> writes:
>> .  the heap tuple header is already formatted, i.e. t_hoff is valid
>
>I think the latter test is unnecessary

Yes, currently.

> and potentially dangerous;

That's the reason, I explicitly mentioned it.

>it could break code that tries to access OID in a
>not-yet-completely-built header.  The test that insists t_hoff is valid
>before one can *set* OID is even more likely to cause trouble.

This was actually the case in one or two places;  I had to change the
order of assignments there.  The code does not contain too many places
where a heap tuple header is built.  If the assertions break the code,
we (or in this case: I) have to fix it.  I need the assertions to find
those spots, if there are any left.

>I do not see the point of this in any case.

I see you have read my proposal on -hackers in the meantime.  Does
that make it clearer?

>            regards, tom lane
Thanks for reviewing my patch!

Servus
 Manfred