Re: Fusion-ifying proto parse trees

Eric Niebler wrote:

> Not to worry, we're not stuck in a groove. :-) I asked you why you 
> needed flat sequences because, when I get hung up on a design problem, I 
> often find it useful to go back to the beginning, start from first 
> principles, and make sure that I haven't overlooked something. That's 
> all. I'm not trying to tell you what you do or don't need, I'm just 
> trying to understand the problem.

Understood.

> OK, you're saying the problem is the performance of creating a view. You 
> agree, though, that there's no sense redesigning proto until we do the 
> performance testing that shows there is a problem that is best addressed 
> that way. Otherwise, it's just premature optimization.

Fair enough. Let me remind you though that the competition is tough.
What we are up against (flattening the sequence up front in the
proto-ET stage) is zero-overhead. When you flatten the sequence
up front, you do as much as you can in a single stage, such that
the stage is already prepared when you go to the second; i.e. parsing.
I still believe that that's the way to go.

Anyway, out of curiosity, how do you intend to do performance testing?
What will be your benchmark?

> I'm going to spend some time thinking about segmented Fusion algorithms, 
> because I think that's where we'll get the most bang for our buck.

That would be awesome!

Regards,
--
Joel de Guzman
http://www.boost-consulting.com
http://spirit.sf.net



-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642