Re: [bpel-dev] Questions about the "empty BPEL" template

[Oliver <olly98@xxxxxxxxxxxx>]

Hi,

>> I think, you are assuming that a BPEL process should be a valid BPEL process
>> at all stages. In my oppinion, a BPEL process has to be valid at the point,
>> the modeling ist completed.
> Not all stages, of course. :)
> But at two moments: at the starting and ending points.

The asynchronous BPEL process seems to be valid at first sight, but is
NOT runnable: The variable "output" is not initialized. The empty
indicates that there should something happen... One has to replace the
empty with some business logic...

> To be honest, I don't know any wizard that results in something invalid.

I see your point...

> But for a new user, an initial error is something which is difficult to
> understand. He may think there is a bug or that he did something wrong in
> the wizard. It is a usability issue.

Then I'd propose to just remove the "empty" activity in the template.
Then, it's a really empty process.

Even if that BPEL process is still invalid, the UI should not force
the user to create a BPEL process based on the Wizrd. New -> "BPEL
process file" always opens the wizard. In the role of a professional,
I want to create an empty BPEL process. That should always be
possible!

> Or we could search an empty process that is valid (much
> better).

An empty process is never valid as there is no start activity :-\

> Otherwise, we still have the less-perfect solution: display a warning in the
> wizard when a user picks up the empty template. This way, he knows that what
> he is about to create will be incomplete (normal) and invalid (unusual).
> This warning would not be displayed if the process is abstract.

100% agree!

Cheers,

Oliver