On Thu, 24 Aug 2000, Rickard [iso-8859-1] �berg wrote:
> ...
> One thing that needs to be done is pooling of prepared statements. That
> should easily be added to Minerva I think. Aaron, any thoughts?
Sure but... Minerva treats everything in the pool as identical,
so you would have to set up different pools for statements with different
SQL. Not a biggie. Also, you'd have to keep the Connections that created
the PreparedStatements out of the pool - since you wouldn't want someone
else committing/rolling back - and this could be a resource drain on the
connection pool. Also, we'd have to fix the problem with keeping a
connection checked out during commit/rollbacks - that's pretty easy, but I
want to be sure we get the theory right first. Finally, the database
should really do this for you. That is, it should cache the query
analysis so you should get most of the speedup by simply using prepared
statements, not by caching them. I'm pretty sure Oracle works this way,
though I don't know that much about the implementations of Hypersonic,
Postgres, etc.
So I think it's possible, but I think the bang-for-buck is pretty
low, particularly when you start using a large number of prepared
statements and beans and you're limited to 255 database connections or
whatever.
Aaron
--
--------------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Problems?: [EMAIL PROTECTED]
