Re: heap_create with OID? - Mailing list pgsql-hackers

From Philip Warner
Subject Re: heap_create with OID?
Date
Msg-id 3.0.5.32.20000704174343.02546410@mail.rhyme.com.au
Whole thread Raw
In response to Re: heap_create with OID?  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: heap_create with OID?
List pgsql-hackers
At 02:57 4/07/00 -0400, Tom Lane wrote:
>Philip Warner <pjw@rhyme.com.au> writes:
>> I am wondering if anyone can see an immediate problem in creating a
function:
>>     heap_create_oid
>> which is just like heap_create, but takes an oid arg. This oid (assuming
>> it's free) would be used to create the object.
>
>How exactly do you propose to determine that the OID is free?

Based on the comments so far, I don't.

pg_dump already allows a 'with-oid' option for table dumping & loading, and
the way it gets around the problem is to save the max oid at dump time,
then set the max oid at restore time. I am only seeking a way of doing a
valid backup/restore which includes BLOBS; so even if I say to the user
that BLOBs can only be restored as part of a full restore, then I think
it's a lot better than the current system.


>There's no inherent problem in forcing a particular choice of OID;
>initdb does it on every run.  The trick is to avoid a conflict of
>OID assignments.

OK. This is good news. The way pg_dump sets the max oid is:
   CREATE TABLE pgdump_oid (dummy int4);   COPY pgdump_oid WITH OIDS FROM stdin;   1282249 0   \.

where 1282249 is the max oid you want to set.

It's pretty crude, but I assume effective. What would be very nice is if I
could 'reserve' the OIDs of the to-be-created BLOBS without having to
create rows/tables etc, and without affecting the max oid. But that is
probably pointless if a decent BLOB implementation is coming along.


>My own feeling is that the current LO setup is fundamentally flawed

You will get no disagreement from me.


>the right answer is
>to find a way to avoid that.

I'm reluctant to do too much in this area until the capabilities of TOAST
are sorted out...

ISTM that, even if a new BLOB architecture comes along, some kind of
migration utility will be needed.

Does this all sound reasonable?

----------------------------------------------------------------
Philip Warner                    |     __---_____
Albatross Consulting Pty. Ltd.   |----/       -  \
(A.C.N. 008 659 498)             |          /(@)   ______---_
Tel: (+61) 0500 83 82 81         |                 _________  \
Fax: (+61) 0500 83 82 82         |                 ___________ |
Http://www.rhyme.com.au          |                /           \|                                |    --________--
PGP key available upon request,  |  /
and from pgp5.ai.mit.edu:11371   |/


pgsql-hackers by date:

Previous
From: Philip Warner
Date:
Subject: Re: heap_create with OID?
Next
From: Philip Warner
Date:
Subject: Re: [GENERAL] Re: proposed improvements to PostgreSQL license