Re: [ebxml-dev] questions on the BPSS spec

[Nandini Ektare <nandini.ektare@sun.com>]

Hi Yamada,

Please find my comments inline....

-Nandini

yakeishi@mail.goo.ne.jp wrote:

> Hi all,
>
> I have three questions on the BPSS.
>
>  1. I cannot find the semantics of the four status values;
>
>     Success, BusinessFailure, TechnicalFailure, and AnyFailure.
>
>     Can anyone show me the description and some example?

I guess Success is fairly easy to judge. If there are none of the three
failures, its success.
The other three failures provide a mechanism to classify different
exceptions that can occur. I think it will be implementation specific as
to how to classify an exception under these categories.(The spec however
helps in classifying certain exceptions as good candidates for 'Business
Failures' such as Timeouts/Business Protocol exceptions.) What the spec
stresses upon is 'unambiguity and finiteness of a business transaction
when any of these failures occur'.

>  2. It seems vague how the status is determined after a
>
>     Collaboration Activity is executed. I guess it is the final
>
>     state of the nested Binary Collaboration. Is this right?

That's right.

>  3. What is the scope of the name attributes?
>
>     One possible and simple interpretation is a name must be
>
>     unique in one ProcessSpecification through the all element
>
>     types (like as XML ID value must be unique even if attribute
>
>     names are different.); e.g., an InitiatingRole and a
>
>     BusinessDocument cannot have the same name even their element
>
>     types are different.
>
>     But the spec doesn't tell me whether it is right.

I feel this simple interpretation may fail in case of 'Includes'. A
ProcessSpec can include another and if IDs are unique within a certain
process spec, during inclusion, there may be a clash in these names. I
feel a better way would be to construct a name of a element similar to a
dn in a ldap tree. For example BTA could be qualified as
ProcSpecID:BCID:BTAID whereas  a resuable element like BT can be
ProcSpecID:PKGID:BTID or a reusable BD element can be ProcSpecID:BDID.
Construction of an ID is again an implementation specific issue.

> I will be happy if anyone can shed some light.
>
> Thanks in advance,
>
> --
>
> Yamada Takafumi
>
> yakeishi@mail.goo.ne.jp
>
> ----------------------------------------------------------------
> The ebxml-dev list is sponsored by OASIS.
> To subscribe or unsubscribe from this elist use the subscription
> manager: <http://lists.ebxml.org/ob/adm.pl>