RE: Questions re ODE2.0-beta

[Dhanush Gopinath <dh...@altair.com>]

Hi,

Is point 2 below happening when you are using initiate with a value
"join"? I couldn't deploy a process which had correlation set with
initiate as "join", but it worked with initiate as "yes". 

Thanks 
Dhanush

-----Original Message-----
From: Tammo van Lessen [mailto:tvanlessen@gmail.com] 
Sent: Thursday, January 28, 2010 2:50 AM
To: user@ode.apache.org
Subject: Re: Questions re ODE2.0-beta

Hi Dawg,

On 27.01.2010 14:56, dawg wrote:
> 1. The instance management implementation of fault() is empty at the
moment.
> Is there a plan to implement it? 

Since it is part of the API it should be implemented. However AFAIK
nobody is currently working on that. If you'd want to create and submit
a patch, that would be highly appreciated ;)

> 2. I could not get correlation to work with 2.0. Is there a known
problem
> with it or is the problem local for me only?

A couple of weeks ago I've fixed an issue related to correlation. Please
see ODE-721, perhaps it's the root cause for your problems, too?

> 3. Implicit initialisation for variables - this is one of the more
annoying
> side effects of using ODE comparing to other bpel engines. Is there a
plan
> to enhance this behaviour? (I know its part of the bpel spec, but it
makes
> the development of complex processes quite hard as you have to
explicitly
> initialise all variables before you can use them).

I tend to agree but the problem is that the initialization mechanism
will never be complete as it is impossible to guess if or how many
optional elements/attributes should be placed in the variable. Hence, I
really think this issue should be properly tackled by BPEL designers,
not by the engine itself. However, if someone is willing to provide
concept and implementation for such a feature, I'd be the last one who
would object to incorporate into ODE (as long as it does not break the
standard behavior).

Best,
  Tammo

-- 
Tammo van Lessen - http://www.taval.de

RE: Questions re ODE2.0-beta

[dawg <zo...@gmail.com>]

No, it happened with "yes". I'm using the distribution by the way, didn't
build the latest version to check with it. 


Dhanush Gopinath wrote:
> 
> Hi,
> 
> Is point 2 below happening when you are using initiate with a value
> "join"? I couldn't deploy a process which had correlation set with
> initiate as "join", but it worked with initiate as "yes". 
> 
> Thanks 
> Dhanush
> 
> -----Original Message-----
> From: Tammo van Lessen [mailto:tvanlessen@gmail.com] 
> Sent: Thursday, January 28, 2010 2:50 AM
> To: user@ode.apache.org
> Subject: Re: Questions re ODE2.0-beta
> 
> Hi Dawg,
> 
> On 27.01.2010 14:56, dawg wrote:
>> 1. The instance management implementation of fault() is empty at the
> moment.
>> Is there a plan to implement it? 
> 
> Since it is part of the API it should be implemented. However AFAIK
> nobody is currently working on that. If you'd want to create and submit
> a patch, that would be highly appreciated ;)
> 
>> 2. I could not get correlation to work with 2.0. Is there a known
> problem
>> with it or is the problem local for me only?
> 
> A couple of weeks ago I've fixed an issue related to correlation. Please
> see ODE-721, perhaps it's the root cause for your problems, too?
> 
>> 3. Implicit initialisation for variables - this is one of the more
> annoying
>> side effects of using ODE comparing to other bpel engines. Is there a
> plan
>> to enhance this behaviour? (I know its part of the bpel spec, but it
> makes
>> the development of complex processes quite hard as you have to
> explicitly
>> initialise all variables before you can use them).
> 
> I tend to agree but the problem is that the initialization mechanism
> will never be complete as it is impossible to guess if or how many
> optional elements/attributes should be placed in the variable. Hence, I
> really think this issue should be properly tackled by BPEL designers,
> not by the engine itself. However, if someone is willing to provide
> concept and implementation for such a feature, I'd be the last one who
> would object to incorporate into ODE (as long as it does not break the
> standard behavior).
> 
> Best,
>   Tammo
> 
> -- 
> Tammo van Lessen - http://www.taval.de
> 
> 

-- 
View this message in context: http://old.nabble.com/Questions-re-ODE2.0-beta-tp27339464p27439379.html
Sent from the Apache Ode User mailing list archive at Nabble.com.