You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@avalon.apache.org by Berin Loritsch <bl...@apache.org> on 2001/12/28 20:16:47 UTC

[VOTE] Oficial Policy For Component Interfaces

Concidering the fact that most Avalon systems automatically determine the lifecycles
of the components, I am wondering if we should strive to maintain 100% backwards
compatibility for lifecycle interfaces.  The issue is brought to light due to the
LogEnabled interface.



Should it be concidered backwards compatible for a *Component* to change it's lifecycle
interfaces?

VOTE HERE:


PLEASE NOTE:
This does not apply to regular classes and containers.  This is only for Components,
which typically have automated mechanisms for starting them up.

-- 

"They that give up essential liberty to obtain a little temporary safety
  deserve neither liberty nor safety."
                 - Benjamin Franklin


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: [VOTE] Oficial Policy For Component Interfaces

Posted by Peter Donald <pe...@apache.org>.
On Sat, 29 Dec 2001 06:16, Berin Loritsch wrote:
> Concidering the fact that most Avalon systems automatically determine the
> lifecycles of the components, I am wondering if we should strive to
> maintain 100% backwards compatibility for lifecycle interfaces.  The issue
> is brought to light due to the LogEnabled interface.
>
>
>
> Should it be concidered backwards compatible for a *Component* to change
> it's lifecycle interfaces?

It would be nice if the contract was at the work interface and lifecycle 
interfaces being completely othogonal to this could easily change etc. If you 
are automagically running the container via ECM, Phoenix or similar then it 
would have a net effect of 0%

However this is not always the case and sometimes you end up hand managing 
components for whatever reason. So it would need to be done on a case by case 
basis. In Phoenix the only way that Blocks talk to each other is via their 
work interface so it would be fine there ... wish ECM did things like this ;)

I don't use any of the clusters like component.* so I don't have an opinion 
whether it is useful to change them. I would prefer that you just added a new 
class ECM2 or maybe something with a better name ;] that implemented all new 
changes and deprecated the old cluster.

-- 
Cheers,

Pete

--------------------------------
 These aren't the droids you're 
 looking for. Move along. 
--------------------------------


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: [VOTE] Oficial Policy For Component Interfaces

Posted by Peter Royal <pr...@managingpartners.com>.
On Friday 28 December 2001 03:50 pm, you wrote:
> I am saying that *strict* backwards compatibility would not allow me to
> change the implementation of a Component from Loggable to LogEnabled.
> If we _relax_ that requirement so that only the role interface and
> contracts surrounding it must remain backwards compatible, then I could
> change Loggable to LogEnabled.  I would also be able to later add
> Initializable, or remove Contextualizable.
>
> For Components, the lifecycle interfaces are supposed to be different
> concerns.
>
> so the vote is: Do we enforce *strict* backwards compatibility or *relaxed*
> backwards compatibility?

(although my vote doesn't count)... I say relax it. If you are using a 
component you should always be taking it through the full lifecycle 
initialization, and if you have a custom container that doesn't do so, too 
bad so sad.

Of course with some of the code that you are talking about creating as part 
of the ECM overhaul (the lifecycle helper) sounds like it will make creating 
a custom container much easier.
-pete

-- 
peter royal -> proyal@managingpartners.com

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: [VOTE] Oficial Policy For Component Interfaces

Posted by Eung-ju Park <co...@apache.org>.
----- Original Message -----
From: "Paul Hammant" <Pa...@yahoo.com>
To: "Avalon Developers List" <av...@jakarta.apache.org>
Sent: Saturday, December 29, 2001 6:25 AM
Subject: Re: [VOTE] Oficial Policy For Component Interfaces


> Berin,
>
> > so the vote is: Do we enforce *strict* backwards compatibility or
> > *relaxed*
> > backwards compatibility?
>
> +1 for Relaxed

+1 for relaxed.

>
> Peter Donald is still on a break though.....
>
> Regards,
>
> - Paul H
>
>
>
> --
> To unsubscribe, e-mail:
<ma...@jakarta.apache.org>
> For additional commands, e-mail:
<ma...@jakarta.apache.org>
>
>


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: [VOTE] Oficial Policy For Component Interfaces

Posted by Paul Hammant <Pa...@yahoo.com>.
Berin,

> so the vote is: Do we enforce *strict* backwards compatibility or 
> *relaxed*
> backwards compatibility? 

+1 for Relaxed

Peter Donald is still on a break though.....

Regards,

- Paul H



--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: [VOTE] Oficial Policy For Component Interfaces

Posted by Berin Loritsch <bl...@apache.org>.
Paul Hammant wrote:

> Berin,
> 
>>
>> Either way the effect is the same.
>>
>> Should I have the ability to change a lifecycle interface without it 
>> being
>> concidered backwards incompatible.
> 
> 
> Sorry dude, I'm still not quite there.  It is not so much of a 
> declaration as an ambiguity.  This from me being the worst master of the 
> English language in this mail list.
> 
> Is the declaration? : We will try not change any of the lifecycle 
> methods from now on for Components.  If we do, we *will* maintain 
> backwards compatibility.



I am saying that *strict* backwards compatibility would not allow me to
change the implementation of a Component from Loggable to LogEnabled.
If we _relax_ that requirement so that only the role interface and
contracts surrounding it must remain backwards compatible, then I could
change Loggable to LogEnabled.  I would also be able to later add
Initializable, or remove Contextualizable.

For Components, the lifecycle interfaces are supposed to be different
concerns.

so the vote is: Do we enforce *strict* backwards compatibility or *relaxed*
backwards compatibility?

-- 

"They that give up essential liberty to obtain a little temporary safety
  deserve neither liberty nor safety."
                 - Benjamin Franklin


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: [VOTE] Oficial Policy For Component Interfaces

Posted by Paul Hammant <Pa...@yahoo.com>.
Berin,

>
> Either way the effect is the same.
>
> Should I have the ability to change a lifecycle interface without it 
> being
> concidered backwards incompatible.

Sorry dude, I'm still not quite there.  It is not so much of a 
declaration as an ambiguity.  This from me being the worst master of the 
English language in this mail list.

Is the declaration? : We will try not change any of the lifecycle 
methods from now on for Components.  If we do, we *will* maintain 
backwards compatibility.

If you had to ask me (again Excalibur is not my area). I would say do 
not bother worrying.  We have less than ten publicly known about 
projects based on Avalon.  There are less than 30 in total I guess.  If 
we were talking about the servlet API there would be war if backwards 
compatibility were not maintained.

Regards,

- Paul




--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: [VOTE] Oficial Policy For Component Interfaces

Posted by Berin Loritsch <bl...@apache.org>.
Paul Hammant wrote:

> Berin,
> 
>> Concidering the fact that most Avalon systems automatically determine 
>> the lifecycles
>> of the components, I am wondering if we should strive to maintain 100% 
>> backwards
>> compatibility for lifecycle interfaces.  The issue is brought to light 
>> due to the
>> LogEnabled interface.
>>
>>
>>
>> Should it be concidered backwards compatible for a *Component* to 
>> change it's lifecycle
>> interfaces? 
> 
> 
> Err sorry to be pedandtic, but do you mean ...
> 
>    Should it be a goal for Components to maintain backwards 
> compatibility of their lifecycle interfaces?
> 
> OR
> 
>   Is the changing of lifecycle interfaces considered to be backwards 
> compatible?


Either way the effect is the same.

Should I have the ability to change a lifecycle interface without it being
concidered backwards incompatible.


> 
>> This does not apply to regular classes and containers.  This is only 
>> for Components,
> 
> 
> And only, mostly to Excalibur I guess.
> 
> Regards,
> 
> - Paul H
> 
> 
> -- 
> To unsubscribe, e-mail:   
> <ma...@jakarta.apache.org>
> For additional commands, e-mail: 
> <ma...@jakarta.apache.org>
> 
> .
> 



-- 

"They that give up essential liberty to obtain a little temporary safety
  deserve neither liberty nor safety."
                 - Benjamin Franklin


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: [VOTE] Oficial Policy For Component Interfaces

Posted by Paul Hammant <Pa...@yahoo.com>.
Berin,

> Concidering the fact that most Avalon systems automatically determine 
> the lifecycles
> of the components, I am wondering if we should strive to maintain 100% 
> backwards
> compatibility for lifecycle interfaces.  The issue is brought to light 
> due to the
> LogEnabled interface.
>
>
>
> Should it be concidered backwards compatible for a *Component* to 
> change it's lifecycle
> interfaces? 

Err sorry to be pedandtic, but do you mean ...

    Should it be a goal for Components to maintain backwards 
compatibility of their lifecycle interfaces?

OR

   Is the changing of lifecycle interfaces considered to be backwards 
compatible?

> This does not apply to regular classes and containers.  This is only 
> for Components,

And only, mostly to Excalibur I guess.

Regards,

- Paul H


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>