← Back to team overview

drizzle-discuss team mailing list archive

Re: Improvements to Storage Engine API

 

Paul McCullagh wrote:

On Jan 4, 2010, at 3:54 AM, Stewart Smith wrote:

On Thu, Dec 24, 2009 at 12:19:13PM +0100, Paul McCullagh wrote:
It is all or nothing, but this done _not_ exclude "partial"
solutions such as condition push-down and sub-select execution.

Would this better be served by stackable parsers? An engine could just
attach a parser and do all the work itself.. as otherwise you're
parsing things twice, so it'll never perform well.

Yes, performance is something I have also been thinking about.

In general, while it would be great to have a GPB representation of DML, I am wondering about the performance cost of building this in addition to everything else (AST, query plan, etc).

In the long run, I don't think we should be passing anything that is not a GPB in the storage engine API except possibly a reference to the current Session object. GPB messages can be constructed from anything that is now a struct or class (LEX, JOIN, etc...) The advantage to a GPB message is clearly in documentation, versioning and serialization of course.

In the case of DDL, this is already required for replication, so this is no additional cost.

Correct.

But, since Drizzle uses RBR a GPB representation of DML is not required.

Well, SELECT is also DML...

-jay



Follow ups

References