Forum OpenACS Development: lob hack in nspostgres and OpenACS

Collapse
1: lob hack in nspostgres and OpenACS
Posted by Jamie Rasmussen on 02/22/03 05:14 PM
Have lobs in PostgreSQL advanced to the point where we could remove the lob hack in nspostgres and OpenACS?
Collapse
2: Re: lob hack in nspostgres and OpenACS (response to 1)
Posted by Don Baccus on 02/23/03 03:27 AM
Not really ... have they actually advanced?  :)

My main problem with them is that you can't use text output from pg_dump, only one of the binary options, and if pg_dump hoses something that means you can't go in and hand-edit it to the point where you can reload the dump.

That scares me ... I've never lost data due to PG_DUMP messing up but being able to hack the text output gives me a comfortable feeling.

Also the LOB API is really clunky, IIRC they're not VACUUMed etc.

Collapse
3: Re: lob hack in nspostgres and OpenACS (response to 2)
Posted by Jamie Rasmussen on 02/23/03 02:01 PM
That's what I suspected.  I was just hoping we could boost performance and open up nspostgres's blob functions to non-OpenACS uses.  It looks like we'll have to wait until those items get checked off of the PostgreSQL TODO list.  Thanks!
Collapse
4: Re: lob hack in nspostgres and OpenACS (response to 1)
Posted by Lamar Owen on 03/05/03 03:08 AM
Last I checked, bytea was toasted and not a 'large object'.

The issue is formatting the data in and out properly in queries.

But bytea's are dumped and vacuumed, if memory serves.  Straight text type is just a special case of bytea.  I'll have to check around....

Collapse
5: Re: lob hack in nspostgres and OpenACS (response to 1)
Posted by Lamar Owen on 03/05/03 03:12 AM