Re: svn commit: r693059 [1/2] - in /myfaces/core/trunk_1.2.x/impl/src/main/java/org/apache/myfaces: application/ application/jsp/ context/portlet/ context/servlet/ convert/ el/ el/convert/ el/unified/resolver/ renderkit/html/ util/

[Simon Kitching <sk...@apache.org>]

Hi Bernd,

bommel@apache.org schrieb:
> Author: bommel
> Date: Mon Sep  8 04:42:28 2008
> New Revision: 693059
>
> URL: http://svn.apache.org/viewvc?rev=693059&view=rev
> Log:
> (MYFACES-1956) Minor performance improvements for org.apache.myfaces.el, org.apache.myfaces.context and org.apache.myfaces.util ...
>
> Apply many of the findbugs and pmd performance rules
>   

Changing classes and methods to be final will break any existing user 
code that subclasses it. Allowing users to subclass stuff inside the 
myfaces impl jar seems rather useful to me. Do you have some performance 
measurements that show this kind of lockdown of the core impl code is 
worthwhile? Or is there some reason why users would never want to 
subclass these packages?

Making a method-parameter final seems less problematic. I'm not quite 
sure what the benefit is though; I would have thought that a decent 
compiler would figure that out itself 99% of the time. Do you have any 
information on what gains we get from this?

Regards, Simon

Re: svn commit: r693059 [1/2] - in /myfaces/core/trunk_1.2.x/impl/src/main/java/org/apache/myfaces: application/ application/jsp/ context/portlet/ context/servlet/ convert/ el/ el/convert/ el/unified/resolver/ renderkit/html/ util/

[Bernd Bohmann <be...@atanion.com>]

Hi Simon,

for performance data please look at

https://issues.apache.org/jira/browse/MYFACES-1956

I don't see any reason why anybody should extend classes from
myfaces-impl. I think jsf has enough extensions points. And if someone
need these impls, we should provide these parts in myfaces-commons.

Regards

Bernd

Simon Kitching schrieb:
> Hi Bernd,
> 
> bommel@apache.org schrieb:
>> Author: bommel
>> Date: Mon Sep  8 04:42:28 2008
>> New Revision: 693059
>>
>> URL: http://svn.apache.org/viewvc?rev=693059&view=rev
>> Log:
>> (MYFACES-1956) Minor performance improvements for
>> org.apache.myfaces.el, org.apache.myfaces.context and
>> org.apache.myfaces.util ...
>>
>> Apply many of the findbugs and pmd performance rules
>>   
> 
> Changing classes and methods to be final will break any existing user
> code that subclasses it. Allowing users to subclass stuff inside the
> myfaces impl jar seems rather useful to me. Do you have some performance
> measurements that show this kind of lockdown of the core impl code is
> worthwhile? Or is there some reason why users would never want to
> subclass these packages?
> 
> Making a method-parameter final seems less problematic. I'm not quite
> sure what the benefit is though; I would have thought that a decent
> compiler would figure that out itself 99% of the time. Do you have any
> information on what gains we get from this?
> 
> Regards, Simon
> 
>