From: | Joe Conway <mail(at)joeconway(dot)com> |
---|---|
To: | Thomas Hallgren <thhal(at)mailblocks(dot)com> |
Cc: | Peter Eisentraut <peter_e(at)gmx(dot)net>, "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org>, bryan(at)bulten(dot)ca |
Subject: | Re: Status of server side Large Object support? |
Date: | 2004-11-28 18:50:16 |
Message-ID: | 41AA1DE8.9050202@joeconway.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Thomas Hallgren wrote:
> Peter Eisentraut wrote:
>> Am Sonntag, 28. November 2004 12:33 schrieb Thomas Hallgren:
>>
>>> Hmm, ok. But there's no way to stream them in and out from disk. From
>>> what I can see, you have to bring all of it into memory. Not so ideal
>>> perhaps if you want to provide streaming media for thousands of users.
>>
>> You can use the substring function to read the pieces you need.
>>
> Won't the substring function bring the whole thing into memory in the
> backend before it pass you the piece you need? Let's assume I want to
> stream 4k at a time of a 40mb image, that's a whole lot of byte swapping
> if that's the case.
Not if the column is storage type EXTERNAL. See a past discussion here:
http://archives.postgresql.org/pgsql-general/2003-07/msg01447.php
> How do you handle writes without first creating the whole image in memory?
You can't currently, but it would be a nice addition ;-)
I agree with Peter -- I think effort is better spent improving bytea.
BTW, someone on GENERAL just started a very similar thread
("implementing a lo_truncate operation, and possibly a lo_size"). I've
cc'd him here.
Joe
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2004-11-28 18:52:24 | Re: Stopgap solution for table-size-estimate updating problem |
Previous Message | Dave Page | 2004-11-28 18:27:01 | Re: Error: column "nsptablespace" does not exist |