Re: Column Sizes - Mailing list pgsql-general

From Adam Ruth
Subject Re: Column Sizes
Date
Msg-id E973E642-15FA-11D8-A573-000A959D1424@intercation.com
Whole thread Raw
In response to Re: Column Sizes  ("Brett Maton" <brett.maton@itaccounting.co.uk>)
Responses Re: Column Sizes
List pgsql-general
On Nov 13, 2003, at 2:47 AM, Brett Maton wrote:

> Thanks for the answers guys,
>
>   I ran a quick query lastnight select length(data) from pg_largeobject
> where loid = xxxx  which prompted my post.
>
>   The query returned 'n' rows of 2048 bytes and an incomplete "page"
> of say
> 236, as I wasn't really thinking about it that much I assumed for some
>   reason that  the data column was limited to a maximum of 2048 bytes
> by the
> database.  I was hoping to select this value (2048) from a query
>   (like \d tablename describes the columns and sizes) to build a buffer
> either a) large enough or b) sensibly proportioned.
>
>   Although my question may not be relevant to this particular issue
> there
> are times that I would like to now the size of a particular column,
> and I
> believe
>   that it's possible to get this value from the database using a query
> and
> pg_catalog etc, just not sure how to achieve this.
>
> Thanks again,
>
> Brett
>
>
> "Brett Maton" <brett.maton@itaccounting.co.uk> wrote in message
> news:botsl3$tsp$1@news.hub.org...
>> Hi NG,
>>
>>   How do I find out the size of a column ?
>>
>>   I am retrieving large objects from the pg_largeobject table and
>> creating
> a
>> files, I would like to know the length of the data column so that I
>> can
>> implement buffering instead of writing hundreds of small chunks to
>> disk.
>>
>> Any help would be appreciated.
>>
>> Thanks in advance,
>>
>>  Bret

The column type for pg_largeobject.data is bytea with no length
specified.  That means, unfortunately, that there's no way to get the
size of the column from the catalog.  It does seem that the large
object functions do split the data into 2k chunks, but it would
probably be a simple thing to check the source to determine if this is
always true.  But it could always change in a future version.

Adam Ruth


pgsql-general by date:

Previous
From: Jan Wieck
Date:
Subject: Re: [HACKERS] Proposal for a cascaded master-slave replication system
Next
From: Rajesh Kumar Mallah
Date:
Subject: DOMAIN usability