sarah,

that is very helpful.  i will experiment with all of your solutions.

i am wondering if (after employing the other performance techniques you
suggested) would the following would also be possible:

imagine the following scenario... 

instead of building 1 large app, i create two smaller lazslo apps that
can be hosted in the same web page and alternately shown and hidden
using dhtml and javascript.  

would that be faster than one larger app?  are two small swf's able to
perform faster than one larger one?  or would that not make any
difference.

obviously, i am imagining extending that model even further so that a
single web page could host maybe 5 or 6 lazslo apps that can be visually
swapped out.  memory consumption would be the same (or worse), but might
performance be superior for some other reason?

paul










-----Original Message-----
From: Sarah Allen [mailto:[EMAIL PROTECTED] 
Sent: Friday, February 17, 2006 7:43 PM
To: Paul Younger
Cc: [email protected]
Subject: Re: [Laszlo-user] Runtime Performance


Laszlo Mail is a large application that performs well. Performance 
doesn't come for free -- that's just the reality of these applications, 
whether they're done with Laszlo or AJAX. The good news is that there 
are many techniques and APIs that allow you to performance-tune a Laszlo

app.

For example: lazy replication reduces the number of views required for 
displaying replicated data; initstage attribute can control when part of

the application instanciate; and dynamically loaded libraries can defer 
download size and load time for less frequently used operations. There 
are also performance measurement tools: a call profiler and a size 
profiler.

While Javascript certainly allows you to write spaghetti code, the OOP 
features of Laszlo (class tag, etc) allow us to create significantly 
better structure than what you could do with plain old Javascript. If 
you keep your app free of debugger warnings, you can see at runtime 
warnings when you have refactored incorrectly if there is anything the 
compiler didn't catch.

Java or C++ provide a lot of features that make it comfortable for 
engineers to build large applications, expecially with new fancy tools 
that help with refactoring; however, in creating GUI applications LZX 
offers real flexibility throughout the process that is much more 
difficult to acheive in other languages.

Hope that helps!

Sarah Allen
Director of Enginering
Laszlo Applications Group

Paul Younger wrote:

> thank you, Michael. your comments are helpful.
>
> could you (or anyone else) provide some parameters for what kind of 
> performance to expect from applications with varying numbers of views?

> what are the practical limits? and is runtime performance something 
> that could be expected to improve significantly in the near future?
>
> another question, how would one expect the runtime performance of the 
> flash-based ui to compare with an ajax-based approach?
>
> and one last question: do you see people sub-dividing larger apps into

> small pieces - separate lzx files - to speed the runtime performance 
> of each?
>
> paul
>
>
------------------------------------------------------------------------
>
> *From:* Michael Pliskin [mailto:[EMAIL PROTECTED]
> *Sent:* Friday, February 17, 2006 4:30 PM
> *To:* Paul Younger
> *Cc:* [email protected]
> *Subject:* Re[2]: [Laszlo-user] Changing a nodes parent
>
> Hello Paul,
>
> The basic problem with large applications in Laszlo is the lack of 
> static compiler checks and thus the lack of easy way to perform 
> refactorings. Another major caveat is its asynchronous nature breaking

> the traditional modality concept and requiring a very specific 
> programming style, sometimes introducing a lot of spaghetti code. 
> These two are the major ones when dealing with large application, at 
> least from my experience. Another problems is that XML-based view 
> replication is too slow for large apps, and in general lots of views 
> can introduce the performance problems as well.
>
>-----------------------------------------------------------------------
-
>
>_______________________________________________
>Laszlo-user mailing list
>[email protected]
>http://www.openlaszlo.org/mailman/listinfo/laszlo-user
>  
>


_______________________________________________
Laszlo-user mailing list
[email protected]
http://www.openlaszlo.org/mailman/listinfo/laszlo-user

Reply via email to