[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
Re: [jwt-dev] Open source Eclipse plug-in based on JaWE&Shark
|
Hi,
I fully agree that JWT should get integrated with other Eclipse projects
(Birt, STP, Mylyn). This is not currently a priority, but this is
something that we keep in mind (as any Eclipse committer must do). It
will probably come with time and growing community...
I also agree that JaWE or YAPROC are better tools for Shark than JWT.
JWT tries to be useful and efficient for non-developer business analyst
and for several process engines (it resolves the "Babel Tower" issue of
the BPM [1]). Thus, we cannot introduce into JWT capabilities that are
too technical or specific to one workflow engine. The leitmotiv is to
give to business analyst and workflow developer a powerful and easy
tool, that would help encourage them to make a good use of BPM without
constraint about the engine they can use.
By definition of JWT (see the original proposal [2]), using JWT for
Shark would have the same benefit as using JWT for any engine: make
process design easier and more agile, leverage SOA in processes, and a
lot of things that will come with age...
More technically:
The Project entity is actually the Package in XPDL vocabulary. JWT and
XPDL metamodels have a lot of similarity, but they do not use the same
vacabulary, the WorkflowService uses JWT vocabulary. The wiki page is
not up to date, if you want to see the current status of wam api, you
can see sources at [3]
I hope my answers help you to better understand the aim of JWT.
Regards,
Mickael
[1] http://www.eclipsecon.org/summiteurope2007/presentations/ESE2007_JWT.pdf
[2] http://www.eclipse.org/proposals/jwt/
[3]
http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.jwt/wam/jwt-wam-runtime-api/src/org/eclipse/jwt/wam/api/WorkflowService.java?revision=1.2&root=Technology_Project&view=markup
Pablo Beltran a écrit :
Hi,
License issues make all the code unuseful for you. I could change my
license type, but it would not resolve the full problem because the
projects used by YAPROC are LGPL and some of them should be required
by you.
On the other hand, at this point, your draft is in a very early
state. I would put the frontier of your extesion points before the
communication issues to concentrate in what JWT-wam will do (a draft
of your extension points capabilities) because the communication layer
can be solved by vendors easyly and it has not too much impact from a
Eclipse point of view.
Would be nice if JWT-wam (or other JWT subproject) would resolve some
dificulties related with other Eclipse technologies. For example: BIRT
for reporting, Mylyn to filter processes used by Eclipse projects and
include only the relevant for each context, and so on.
I want to apologise if I'm changing the JWT (JWT-wam) scope:) but I
think on JWT as a bridge between workflows and the entire Eclipse
platform making easier adopt the Eclipse provided techonolgies for
workflow vendors.
Now I'm asking to myself why are (or would be) the benefits of ussing
JWT for Shark. And I can't find a good answer to that question since
Shark has resolved in a gracefuly way the main communication issues
thus it don't get a clear benefit from using JWT. Maybe things are
different for Bonita since JWT is a good exercise to create a new
communication API.
Finally I've a question about your current data model for JWT-wam:
* What does mean the Project entity? It seems like projects belong to
processes.
Best regards,
Pablo.
2009/1/26 Mickael Istria <mickael.istria@xxxxxxxxxxx
<mailto:mickael.istria@xxxxxxxxxxx>>
Hello Pablo,
Thanks for your interest to JWT and for your advices.
I'll try to spend some time today to take a look at YAPROC to see
which interesting features it contains and give you some feedback.
Here are a few comments about JWT that may be interesting for you
to know:
* Since JWT is an Eclipse.org project, we are not allowed to
include into JWT some code or libraries that are under the LGPL or
GPL. Only EPL code is allowed on Eclipse.org code base, and we can
use some code/libs under the BSD, Apache, MIT-like licenses, but
it requires the approval by the Eclipse PMC (if you are interested
in Eclipse Intellectual Property policy, take a look at this page
=> http://www.eclipse.org/legal/EclipseLegalProcessPoster.pdf).
<http://www.eclipse.org/legal/EclipseLegalProcessPoster.pdf%29.>..
* JWT-wam (workflow administration and monitoring) is currently
only a draft. You can find a few details on the page
http://wiki.eclipse.org/JWT_Monitoring. As you can see, we have a
kind of "abstraction layer" that contains some beans to describe
the workflow state, and a WorkflowService interface that contains
a set of operations to query the workflow engine about runtime
data. This API is not definitive since it is often refactored, any
comments are welcome ;)
* To link a workflow engine to JWT, the only thing to do is to
implement this WorkflowService interface. We currently have an
implementation for Bonita v3 with webservices (on SCorWare's
private SVN, since we cannot commit it to Eclipse because of
license), one with an "embedded" Bonita v4, but we could also
think about an implementation directly on databases, thanks to
VirtualDB.
Feel free to criticize it if you want to, it is still in progress
and any other opinion will help us!
Regards,
Mickael
Pablo Beltran a écrit :
YAPROC is a new Eclipse (one moth old) plugin that embeds JaWE
& Shark with many nice features like reporting, process viewer
at runtime, activity management, graphical history,... and
compatible with the XPDL 1.0 specification.
Now, YAPROC becomes an open source project under the LGPL v3
license terms. Maybe you can re-use some parts in your project.
The plugin, documentation, sources and online demos are
available from www.upversion.com <http://www.upversion.com>
<http://www.upversion.com/>
Likely a very interesting feature for you is the VirtualDB
that allows you to integrate differents java technologies into
a jdbc driver. It's not a technology from me but I adapted it
to be successfully used to integrate Shark into a new complex UI.
Based in my experience with Shark (which has a quite complex
API) and YAPROC (which has a quite complex UI) I would
separate UI issues from communication workflow issues. Mybe
you have done this exercise but If not I would like alert to
you about potential design limits to build a complex and
usable UI based on a workflow communication layer when
required functionality grows and grows.
I think now is too soon to implement this kind of solutions in
JWT because your actual UI is quite basic but maybe in the
near future if you develop new plugin views and you need to
relate many of them and many complex filters would be required
to show a lot of information in the Eclipse Workbench in a
coherent manner, so maybe this problem arises and you need to
resolve it. In this case, VirtualDB is a very suitable
technology to solve this kind of problems.
Best regards,
Pablo.
------------------------------------------------------------------------
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx <mailto:jwt-dev@xxxxxxxxxxx>
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx <mailto:jwt-dev@xxxxxxxxxxx>
https://dev.eclipse.org/mailman/listinfo/jwt-dev
------------------------------------------------------------------------
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev