geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Wilkins <gr...@mortbay.com>
Subject Re: [General] Container interface and AbstractContainer
Date Sat, 16 Aug 2003 11:21:55 GMT



Berin Loritsch wrote:
> Dain Sundstrom wrote:
> 
> 
>>
>> BTW, they really aren't Java interfaces, but rather exposed management 
>> interfaces, which are just naming conventions.
>>
> 
> grumble, grumble, snarl, spit...
> 
> <rant type="stupid specs">
>   Why have a language feature like interfaces if you speicify in all your
>   JSRs that everything has to be done by naming conventions and 
> reflection?!
>   I swear that stupid stuff like _forcing_ a naming convention or _forcing_
>   the use of reflection for a tool to work is IMNSHO a pile of dung!

Could not agree more.

In the JSR77 threads we are talking about how to handle this.
We have already created some interfaces for some of the models (State
and StateManageable).    I have suggested that we move these to the
spec/jsr77 module under org.apache.management.j2ee package.

Hopefully Jason will do that.... but if he doesn't and nobody else
complains - then I'll do that.  We should then create interfaces for
all the models that we use from the spec an place them there.

cheers




>   I have been living for quite some time without having to do any of that
>   crap, and I can still get at all my services and components, etc.  Not
>   to mention it operates quicker and is easier to manage and maintain.
> 
>   IMO many of the JSRs out there like this are seriously broken because of
>   all the crap that you have to go through to stay compliant.  Things are
>   so tightly specified that if you ever thought of a better or easier way
>   of doing things you would suddenly be out of spec.
> 
>   One of my complaints about specs like JMX and now JSR077 is that they
>   force the stinking naming convention on you and give you no freedom
>   to improve while remaining true to the intent of the spec.  All it takes
>   is a few extra brain cells and you can solve the problem without the
>   naming convention or braindead implementation.
> </rant>
> 
> Sorry, now that I got that off of my chest, I can help make the broken
> spec work as best possible.  (I know it isn't your fault that's the way
> the spec was written).
> 




Mime
View raw message