In the 1.2 series, we tried to keep memory usage lower by only pulling in
512K of each message into memory at a time, then dumping to the database,
and pulling in the next 512K. This was possible to do at the time because
the hokey mime/header parser only wanted to see that all of the headers
fit into the first 512K block.

GMime has some stream-oriented functionality, but we're primarily using it
in an all-in-memory mode, on the whole message.

Aaron

On Thu, Dec 21, 2006, Matthew O'Connor <[email protected]> said:

> I always thought that messages were split into blocks for MySQL, anyway, 
> bytea fields in PGSQL can support upto 1GB of data.
> 
> Paul J Stevens wrote:
>> That's just for historical reasons.
>> 
>> READ_BLOCK_SIZE is defined in db.h as 512KB, but you can set it to 512MB
>> or even bigger. Mysql supports 4GB for the type in question (longblob),
>> but check the docs if you're using postgresql.
>> 
>> 
>> tom wrote:
>>> What motivated dbmail to split the messages into message_blks?
>>> Is there some insane impracticality to storing a 1MB body as a single
>>> text field?
>>> _______________________________________________
>>> DBmail mailing list
>>> [email protected]
>>> https://mailman.fastxs.nl/mailman/listinfo/dbmail
>>>
>> 
>> 
> _______________________________________________
> DBmail mailing list
> [email protected]
> https://mailman.fastxs.nl/mailman/listinfo/dbmail
> 

-- 



_______________________________________________
DBmail mailing list
[email protected]
https://mailman.fastxs.nl/mailman/listinfo/dbmail

Reply via email to