ObjectWeb Consortium
Search ObjectWeb Mail Archive: 

Advanced Search - Powered by Google

Mail Archive Home | shark List | December 2004 Index

<--  Date Index  --> <--  Thread Index  -->

AW: [shark] A few implementation doubts

Hi Aparna,

I cannot answer all questions you have, but maybe the following is a
possible approach for the first one:

Shark itself does not deliver a native mean to manage files. You need to
implement it yourself, what is pretty simple on top of Shark.

Our use of Shark does exactly what you are requesting, i suppose. I use it
to pass files from one activity to another (or more precise: file names),
where each activity may transform the file in use. Additionally, we store
all file versions - input as well as output - in the file system. I call
these files "stage files", as they represent a stage within the
tranformation. They might also be called "work items" or "artifacts". 

We have written i.e. a xslt transformation activity which takes an input
string for a stage file (content is xml), and generates a new stage file
with the result of the transformation (xml, cvs, text, whatever).

The adiministration of these stage files is simplified by using a utility
class that manages all the versioning stuff in static stateless methods.

I think one advantage of storing the file in the file system, not in a
database, and passing just the stage file names to the activities, is, that
the files can be processed at a very low memory consumption, given the right
parsing engine is used (i.e. sax for xml content). 

Another advantage is, that these stage files represent additional states
belonging to the workflow process instance, so that it can continue where it
left when Shark is restarted. However, I did not test this assumption yet.

Hope this gives you an idea of how this first question can be implemented.

Daniel Frey

-----Ursprüngliche Nachricht-----
Von: Aparna Sivakumar [mailto:aparna@xxxxxxxxxxxxx] 
Gesendet: Dienstag, 21. Dezember 2004 14:43
An: shark@xxxxxxxxxxxxx
Betreff: [shark] A few implementation doubts


In my team, we are trying to develop a business process in which I need to
code the following modules:

1) File flow in the process - document flow from one activity to another (a
File as a Data Type in XPDL that is input/output in the activities) - how
can this be achieved.

2) Mailing with file attachment - Has anyone extended the Mail Tool Agent to
send attachments also in the mail.

3) Just to confirm, right now Shark supports only the basic data types -
String, Integer, Float, Boolean, etc, right?
How does one go about including Java type data inputs - Array, Enumeration,
for instance. 

Any pointers/direction would be extremely helpful.

Best Regards,

Netcore's New Website

<--  Date Index  --> <--  Thread Index  -->

Reply via email to:

Powered by MHonArc.

Copyright © 1999-2005, ObjectWeb Consortium | contact | webmaster.