Oliver Jowett wrote:
> This patch allows bytea parameters set via setBytes() /
> setBinaryStream() to be streamed to the backend. With the patch applied,
> the additional memory overhead for bytea parameters is small and
> independent of how large the parameter data is.
> Note that it doesn't touch the ResultSet path, so you'll still need a
> pile of extra memory if you execute queries that return large bytea values.
> It passes the driver's testcases against a 7.4 server, and doesn't show
> any problems after brief testing with our bytea-using application. I've
> had a report of strange things happening on the server->client path, but
> don't have any details at the moment.
> This needs testing under a real workload, and testing against a 7.3
> server (the V2 protocol path is currently untested). Any volunteers?
I just tried this against a 7.3.6 server and saw no problems (passes the
driver testsuite, boots our application).
In response to
pgsql-jdbc by date
|Next:||From: Moray Taylor||Date: 2004-04-25 20:19:13|
|Subject: Re: Help:updateRow() with CIDR types THANK YOU!|
|Previous:||From: Simon Moses||Date: 2004-04-25 05:13:01|
|Subject: loading data into postgresql became slow after i used VACUUME command.|