Mailing-List: contact flex-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: flex-dev@incubator.apache.org
Received-SPF: pass (athena.apache.org: domain of webdoublefx@hotmail.com
 designates 65.55.111.88 as permitted sender)
Message-ID: <BLU162-ds386534311EDFE7809E98AB48D0@phx.gbl>
From: =?UTF-8?Q?Fr=C3=A9d=C3=A9ric_Thomas?= <webdoublefx@hotmail.com>
To: <flex-dev@incubator.apache.org>
References: 
 <CAJ16SzCMt0732vFM0ieuqp99GQ1GcmfwJpcOGMpVaL4Hha9CKg@mail.gmail.com><CB44F02B.3AEC0%aharui@adobe.com>
 <CAJ16SzDm49WBL2Sj+V0xZK_DaUr8LnqVuNzUDEwUSgae2rBgKg@mail.gmail.com>
 <F85F710414401743B5093C1CCDD0F3AD89984EC463@NAMBX02.corp.adobe.com>
In-Reply-To: 
 <F85F710414401743B5093C1CCDD0F3AD89984EC463@NAMBX02.corp.adobe.com>
Subject: Re: Flex 5 UIComponent - Behavior Pattern
Date: Mon, 30 Jan 2012 01:34:12 +0100
MIME-Version: 1.0
Content-Type: text/plain;
	format=flowed;
	charset="UTF-8";
	reply-type=original
Content-Transfer-Encoding: 8bit
Importance: Normal

Given there is no news on the behavior approach of the UIComponent (I guess 
some of you should still work on it), I did a quick test last night 
splitting the IStateClient2 implememtation of the UIcomponent into a 
StateBehavior, I then ran a test to measure the instatiation time and I've 
been surprised to see that it was quicker than the original UIComponent.


Well, I've been quick and my test is may be not the best to do, I verified 
that the states, transitions and effects worked fine first and then just 
instanciated 100000 UIComponents and compared the time it took relative to 
the original version.
It should may be good for the folks doing the same kind of job on the 
behavior approach at the moment to define a test protocol (time, memory, 
behavior).


May be you can help Alex in that direction, given you already did some tests 
and measurements on it apparently, it would be nice to see your approach, 
what do you think, is it possible ?


Is there someone arrived to the same finding than me ?


#Andres Lozada, I didn't follow the approach you mentioned in my try at the 
moment but it is the one I thought to try too in a full dev.


Frédéric Thomas

-----Message d'origine----- 
From: Alex Harui
Sent: Wednesday, January 25, 2012 11:10 PM
To: flex-dev@incubator.apache.org
Subject: RE: Flex 5 UIComponent - Behavior Pattern

> -----Original Message-----
> From: Haykel BEN JEMIA [mailto:haykelbj@gmail.com]
> Sent: Tuesday, January 24, 2012 11:51 PM
> To: flex-dev@incubator.apache.org
> Subject: Re: Flex 5 UIComponent - Behavior Pattern
>
>
> This looks very bad. It looks like often used features must be
> implemented
> through inheritance.
>
>
Well, that's why I say we have to give up on backward compatibility.  In my 
test, the measurement behavior was one of the most expensive 'behaviors'. 
However, in the current framework, lots of components are unnecessarily 
measured at startup.  We can't change that without breaking backward 
compatibility.  In my whiteboard, I will explore a simpler top-down layout 
that is more HTML-like.

Alex Harui
Flex SDK Developer
Adobe Systems Inc.
Blog: http://blogs.adobe.com/aharui