BLOBS (ODBC lo object) and pg_restore

Jason Godden <jasongodden@optushome.com.au> writes:

./pg_restore -p5431 -dsomedb -Ft -N -X disable-triggers
/var/db/backup/data.tar
ERROR: Relation "BLOBs" does not exist

Looks like a bug in pg_restore. For the moment I'd suggest not using
-X disable-triggers while restoring blobs ...

regards, tom lane

Jason Godden <jasongodden@optushome.com.au> writes:

./pg_restore -p5431 -dsomedb -Ft -N -X disable-triggers
/var/db/backup/data.tar
ERROR: Relation "BLOBs" does not exist

Here's the patch if you need it.

regards, tom lane

*** src/bin/pg_dump/pg_backup_archiver.c~	Sat Feb  1 17:07:14 2003
--- src/bin/pg_dump/pg_backup_archiver.c	Sun Mar  9 14:36:40 2003
***************
*** 293,299 ****
  					 * Maybe we can't do BLOBS, so check if this node is
  					 * for BLOBS
  					 */
! 					if ((strcmp(te->desc, "BLOBS") == 0) && !_canRestoreBlobs(AH))
  					{
  						ahprintf(AH, "--\n-- SKIPPED \n--\n\n");

--- 293,300 ----
  					 * Maybe we can't do BLOBS, so check if this node is
  					 * for BLOBS
  					 */
! 					if ((strcmp(te->desc, "BLOBS") == 0) &&
! 						!_canRestoreBlobs(AH))
  					{
  						ahprintf(AH, "--\n-- SKIPPED \n--\n\n");

***************
*** 445,450 ****
--- 446,455 ----
  	if (!ropt->dataOnly || !ropt->disable_triggers)
  		return;

+ 	/* Don't do it for the BLOBS TocEntry, either */
+ 	if (te && strcmp(te->desc, "BLOBS") == 0)
+ 		return;
+ 
  	oldUser = strdup(AH->currUser);
  	oldSchema = strdup(AH->currSchema);

***************
*** 506,511 ****
--- 511,520 ----

  	/* This hack is only needed in a data-only restore */
  	if (!ropt->dataOnly || !ropt->disable_triggers)
+ 		return;
+ 
+ 	/* Don't do it for the BLOBS TocEntry, either */
+ 	if (te && strcmp(te->desc, "BLOBS") == 0)
  		return;

oldUser = strdup(AH->currUser);

I'd like to select a random record from a table, but I'd like to weight it.

For example, let's say I have a table like:

data1 data2 weight
----------------------------------------
1 2 3.44
3 4 0.94
5 6 1.00
7 8 2.00

I would like to select (data1,data2) from a random record, but I'd like (1,2)
to be 3.44 times as likely as (5,6). I would also like (7,8) to be 2 times as
likely as (5,6), and (1,2) to be 1.72 times as likely as (7,8).

Does that make mathematical sense? I think I can do it if weight is an integer
(but it would be a really bad hack), but I can't think of a nice way with
fractions.

Thanks,
Jeff Davis

On Sun, 9 Mar 2003 13:40:30 -0800, Jeff Davis
<jdavis-pgsql@empires.org> wrote:

I would like to select (data1,data2) from a random record, but I'd like (1,2)
to be 3.44 times as likely as (5,6). I would also like (7,8) to be 2 times as
likely as (5,6), and (1,2) to be 1.72 times as likely as (7,8).

If you had

data1 data2 weight minw maxw
--------------------------------------
1 2 3.44 0.00 3.44
3 4 0.94 3.44 4.38
5 6 1.00 4.38 5.38
7 8 2.00 5.38 7.38

and an immutable wrapper function around random(), you could

SELECT data1, data2
FROM t
WHERE minw < myrandom(7.38) AND myrandom(7.38) <= maxw;

Make sure myrandom() never returns 0.00 or set minw to something less
than 0.00 in the first row.

Servus
Manfred

Hi Tom,

Thanks for this - I'm really impressed with the response. In the end I used
lo_export and PL/PGSQL procedure to dump all the files (60000 of them!) to
disk and I bz2 that up to backup each night. I'll revisit pg_restore/pg_dump
when I get a chance.

I realise that you guys probably have plenty of developers but I'm wanting to
get my C/C++ back up to scratch again and I love Postgres - anything that you
guys need someone to sink their teeth into that I could have a shot at?

Cheers,

Jason