next steps in development of V4RB2? when will be speed a task

[Ruslan Zasukhin]

On 7/29/05 7:36 AM, "Claudius Sailer" <Claudius at sailer-online.de> wrote:

Hi Claudius,
 
> I see V4RB2 is a good environment and a good solution for me and the
> future will bring interesting features ;-)) (#810, #811).

Well, this is not new features. Just more complex SQL queries.

New features which you can expect from Valentina are e.g.
    * Support of collections on kernel level
    * Support of Xquery
    * transactions, hot backup,
    * inheritance of tables.

> But one  point is at the moment badly to me (#641) the speed in a lot of
> circumstances is much slower as in V4RB1.10.
 
> I could speed up my application with V4RB2 at a lot of parts, because  I only
> need one SQL instead of 2 or 3 to come to same result and in a  lot of parts I
> can use now SQL where I hade to write RB-Code to take  correct data and values
> out of database cursor.

> But in my mainpart of  my application. the speed is now 2-3 times slower which
> is not accepted by my users.

> I knew that working on performance and  optimizations will be a part of the
> future (as I understand in this  list correct), but it would be interesting
> when will this start. I  want to be able to plan next releases and steps with
> my application  and it would be nice to have a hint about window of time.

> The good thing is. At the moment I have less of time, so I have no
> problem when this will be not tomorrow ;-))

> My application is running on my computer since 2.0.3 stable without
> any problem. really good.

Yes, really good.
 
> Oh I forgot. It would also be nice when V4RB2 components could be
> part of RB Package

What you mean?  Bundled with REALbasic?

Well, this is unlikely :-)

1) REAL now use SQL Lite,
2) Valentina is fairly big now


----
About speed.

Serge week ago have start develop in REALbasic BenchProject.
This will be public open source project.
I think we will put it into Advanced folder.

This going to be quite complex and big project.

1) it allow us execute single bench under profiler to see where is speed
problems.

2) second main task of this project -- it can execute all existed benches in
all existed variations and put results into database.

This give Paradigma new kind of protection of kernel. Let me explain.
Assume one developer add some new code. We have, as you know about 25,000
tests to check correctness so we can see that we have not broke something.

But there is no guarantee up to now, that we have not kill performance in
some area. Having such auto-benches we will have system that automatically
will warn us if this happens. Then we easy know what changes cause problem.

I hope that such things will allow us ship new versions that are correct and
at least NOT slower than previous builds

WHILE this will be public project everybody can add own bench important for
him then send this to us, so we will check code and add it into project.


-- 
Best regards,

Ruslan Zasukhin
VP Engineering and New Technology
Paradigma Software, Inc

Valentina - Joining Worlds of Information
http://www.paradigmasoft.com

[I feel the need: the need for speed]