Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Eclipse Projects » Java WorkFlow Tooling (JWT) » Castor?
Castor? [message #21844] Sat, 11 February 2006 22:26 Go to next message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

Hi,
I have the project to substitute existing Workflow Designer in Swing with
Castor
(http://www.castor.org) to app in GEF with EMF without touching server-side
BPM engine but providing interoperability with it.

I made questions in relation with it in:
http://www.eclipse.org/newsportal/article.php?id=1670&gr oup=eclipse.technology.xsd#1670

http://www.eclipse.org/newsportal/article.php?id=1667&gr oup=eclipse.technology.xsd#1667

as well as in GEF, EMF; BPEL-DESIGNER Eclipse newsgroups without any clear
answer.


After many discussions in GMF (IMHO the only responding newsgroup) with
culmination in
http://www.eclipse.org/newsportal/article.php?id=1111&gr oup=eclipse.technology.gmf#1111
I have come to some conclusions (I wish I had them before):
- as far as there is no integration priority/intentions inside Eclipse
(extensibility, interoperability, XML transformations/pluggability formats),
it is dangerous to make any dependencies on any other Eclipse frameworks and
projects, even the core ones, like EMF, etc.
- there is some sense to base JWT on a long-established well-approved
non-Eclipse framework. Otherwise JWT is going to overlap/duplicate the work
in GMF, WTP, EMF, WST, VE. STP, BPEL-Deisgner, GMT, MTF, Merlin, EMFT, XSD,
SDO, etc. How about Castor?

BTW, what will be, the equivalent Eclipse frameworks to Castor covering all
its features? and how to cope with the problem of already existing, at
least data, files created with Castor framework?

Guennadi Vanine
Re: Castor? [message #21890 is a reply to message #21844] Sun, 12 February 2006 01:17 Go to previous messageGo to next message
Eclipse User
Originally posted by: allenmacyoung.gmail.com

Yeah, I agree. If what Guennadi said is the fact, I mean things like EMF
are really cannot be integrated together at the projects' level, we should
choose some other framework.

I want to ask why we choose to take this JWT project based on Eclipse. In
my opinion, it's because eclipse platform has its advantages for developping.
But I should also say that not all the things around eclipse are the right
choices. We should make use of eclipse's advantages and combine them with
other excellent frameworks if there is no proper one in eclipse community.

> Hi,
> I have the project to substitute existing Workflow Designer in Swing
> with
> Castor
> (http://www.castor.org) to app in GEF with EMF without touching
> server-side
> BPM engine but providing interoperability with it.
> I made questions in relation with it in:
> http://www.eclipse.org/newsportal/article.php?id=1670&gr oup=eclipse.te
> chnology.xsd#1670
> http://www.eclipse.org/newsportal/article.php?id=1667&gr oup=eclipse.te
> chnology.xsd#1667
>
> as well as in GEF, EMF; BPEL-DESIGNER Eclipse newsgroups without any
> clear answer.
>
> After many discussions in GMF (IMHO the only responding newsgroup)
> with
> culmination in
> http://www.eclipse.org/newsportal/article.php?id=1111&gr oup=eclipse.te
> chnology.gmf#1111
> I have come to some conclusions (I wish I had them before):
> - as far as there is no integration priority/intentions inside Eclipse
> (extensibility, interoperability, XML transformations/pluggability
> formats),
> it is dangerous to make any dependencies on any other Eclipse
> frameworks and
> projects, even the core ones, like EMF, etc.
> - there is some sense to base JWT on a long-established well-approved
> non-Eclipse framework. Otherwise JWT is going to overlap/duplicate the
> work
> in GMF, WTP, EMF, WST, VE. STP, BPEL-Deisgner, GMT, MTF, Merlin, EMFT,
> XSD,
> SDO, etc. How about Castor?
> BTW, what will be, the equivalent Eclipse frameworks to Castor
> covering all its features? and how to cope with the problem of
> already existing, at least data, files created with Castor framework?
>
> Guennadi Vanine
>
Castor+other frameworks, e.g., Cocoon [message #21980 is a reply to message #21890] Sun, 12 February 2006 15:34 Go to previous messageGo to next message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

Thanks, Allen,
following our on-line discussion. I'd like to respond more... what I can.
I am not XML specialist but just mere mortal (if I am not paid I even
shouldn't try to contradict him with my initiatives)

There are around hundred XML parsing and/or marshaling frameworks just
in/for Java (in contrast to simply core System.Xml and System.Xml.Schema in
..NET) and it is hard to assume the role of god in affirming 100% correct
choice

What distinguishes Castor is:
- it is first marshaling framework, docs/samples are abundant and other
frameworks from big players simply did struggle to copy functionalities and
APIs after it;
- flexibility and interoperability with other Open Source frameworks like
XSLT Processors (Xalan, Saxon, etc.) and Cocoon, etc.

My main WISHHHHHH is to have someting like Biztalk Property Promotions
(changing interfaces/contracts, formats on-the-fly),
i.e. to have loose coupling.

And I see how to do it with Castor as a base (possibly in combination of
other frameworks), but not with any other frameworks.

Cf., for ex., CastorTransformer of Apache Cocoon;
http://www.castor.org/projects.html

Guennadi Vanine

"Allen Young" <allenmacyoung@gmail.com> wrote in message
news:a6f703681b198c7fdcc51a2c6c0@news.eclipse.org...
> Yeah, I agree. If what Guennadi said is the fact, I mean things like EMF
> are really cannot be integrated together at the projects' level, we should
> choose some other framework.
>
> I want to ask why we choose to take this JWT project based on Eclipse. In
> my opinion, it's because eclipse platform has its advantages for
> developping. But I should also say that not all the things around eclipse
> are the right choices. We should make use of eclipse's advantages and
> combine them with other excellent frameworks if there is no proper one in
> eclipse community.
>
>> Hi,
>> I have the project to substitute existing Workflow Designer in Swing
>> with
>> Castor
>> (http://www.castor.org) to app in GEF with EMF without touching
>> server-side
>> BPM engine but providing interoperability with it.
>> I made questions in relation with it in:
>> http://www.eclipse.org/newsportal/article.php?id=1670&gr oup=eclipse.te
>> chnology.xsd#1670
>> http://www.eclipse.org/newsportal/article.php?id=1667&gr oup=eclipse.te
>> chnology.xsd#1667
>>
>> as well as in GEF, EMF; BPEL-DESIGNER Eclipse newsgroups without any
>> clear answer.
>>
>> After many discussions in GMF (IMHO the only responding newsgroup)
>> with
>> culmination in
>> http://www.eclipse.org/newsportal/article.php?id=1111&gr oup=eclipse.te
>> chnology.gmf#1111
>> I have come to some conclusions (I wish I had them before):
>> - as far as there is no integration priority/intentions inside Eclipse
>> (extensibility, interoperability, XML transformations/pluggability
>> formats),
>> it is dangerous to make any dependencies on any other Eclipse
>> frameworks and
>> projects, even the core ones, like EMF, etc.
>> - there is some sense to base JWT on a long-established well-approved
>> non-Eclipse framework. Otherwise JWT is going to overlap/duplicate the
>> work
>> in GMF, WTP, EMF, WST, VE. STP, BPEL-Deisgner, GMT, MTF, Merlin, EMFT,
>> XSD,
>> SDO, etc. How about Castor?
>> BTW, what will be, the equivalent Eclipse frameworks to Castor
>> covering all its features? and how to cope with the problem of
>> already existing, at least data, files created with Castor framework?
>>
>> Guennadi Vanine
>>
>
>
Re: Castor? [message #22070 is a reply to message #21890] Mon, 13 February 2006 10:23 Go to previous messageGo to next message
Eclipse User
Originally posted by: fabrice.dewasmes.openwide.fr

Allen Young wrote:
> Yeah, I agree. If what Guennadi said is the fact, I mean things like EMF
> are really cannot be integrated together at the projects' level, we
> should choose some other framework.
>
> I want to ask why we choose to take this JWT project based on Eclipse.
> In my opinion, it's because eclipse platform has its advantages for
> developping. But I should also say that not all the things around
> eclipse are the right choices. We should make use of eclipse's
> advantages and combine them with other excellent frameworks if there is
> no proper one in eclipse community.

Allen, Guennadi,

I like pragmatic choices. We shouldn't use Eclipse frameworks if we feel
this is not the best choice. Let's not take an eclipse framework just to
say we leverage a lot of things in the eclipse area.

the EMF XML generation may be of course limited. If I get you right,
what I understand is that you were wondering about the possible switch
from one representation to another based on the workflow core metamodel
using a transformation engine.

I have read the posts about this in GMF (congrats to Guennadi who has
shot down for spamming ;)). the MTF project seems interesting but not
Open Source. This tags it 'non eligible'. We may look at it when it
comes to the open source community and especially eclipse (which is the
most probable community).

Choosing a transformation technology seems good but which ? To me is
Castor an old school OXM (Object XML Mapping). But I'm also dissatisfied
by others (JAXB and the like)...

Fabrice

>
>> Hi,
>> I have the project to substitute existing Workflow Designer in Swing
>> with
>> Castor
>> (http://www.castor.org) to app in GEF with EMF without touching
>> server-side
>> BPM engine but providing interoperability with it.
>> I made questions in relation with it in:
>> http://www.eclipse.org/newsportal/article.php?id=1670&gr oup=eclipse.te
>> chnology.xsd#1670
>> http://www.eclipse.org/newsportal/article.php?id=1667&gr oup=eclipse.te
>> chnology.xsd#1667
>>
>> as well as in GEF, EMF; BPEL-DESIGNER Eclipse newsgroups without any
>> clear answer.
>>
>> After many discussions in GMF (IMHO the only responding newsgroup)
>> with
>> culmination in
>> http://www.eclipse.org/newsportal/article.php?id=1111&gr oup=eclipse.te
>> chnology.gmf#1111
>> I have come to some conclusions (I wish I had them before):
>> - as far as there is no integration priority/intentions inside Eclipse
>> (extensibility, interoperability, XML transformations/pluggability
>> formats),
>> it is dangerous to make any dependencies on any other Eclipse
>> frameworks and
>> projects, even the core ones, like EMF, etc.
>> - there is some sense to base JWT on a long-established well-approved
>> non-Eclipse framework. Otherwise JWT is going to overlap/duplicate the
>> work
>> in GMF, WTP, EMF, WST, VE. STP, BPEL-Deisgner, GMT, MTF, Merlin, EMFT,
>> XSD,
>> SDO, etc. How about Castor?
>> BTW, what will be, the equivalent Eclipse frameworks to Castor
>> covering all its features? and how to cope with the problem of
>> already existing, at least data, files created with Castor framework?
>>
>> Guennadi Vanine
>>
>
>
Re: Castor? [message #22256 is a reply to message #22070] Mon, 13 February 2006 17:01 Go to previous message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

Hi,
I'd like to resume my previous opinions:
1)
Even if GMT would have been already OpenSrc-ed and integrated to Eclipse:
well, I am still not happy with dependency on EMF.
It is somewhat developer-oriented and OVERKILL for such simple issues as XML
pluggability.
GMF helps with pre-considered during development XML formats/XSD but will
not address extensibility/interoperability even in release (by June, 2006).
As a matter of fact, it even can't approach it since it is EMF and other
frameworks problems/dependence.
1a)
There is no sense to compete with (or make dependencies on) GMF and other
OpenSrc projects if we can go by more simple (but not simpler) directions.
1b)
Eclipse frameworks/libraries are based on pre-fixed contracts/interfaces
Castor + Cocoon can help to break through this dependence also

2)
There is big sense to base base BOTH server-side (BPM engine, WA in JWT
terms?) and client side (workflow designer) on the same framework open to
interoperability and quite simple (but not simpler)!
And how does fit Eclipse frameworks (EMF, GEF, GMF, etc.) into this?

What about adapters, bridging/interoperating with LDAP, databases and other
frameworks that is not the problem in Castor?

3)
JWT should solve original distinguishing problems amongst both other Eclipse
projects and even non-Eclipse OpenSrc projects.
And these are pluggability of XML formats (at client-side, designer) and
loose-coupling at server side

It sounds simple but practically happen to be intricate problem in context
of Eclipse-based projects and have also the conceptual value for non-Eclipse
OpenSrc approaches.

If we base JWT on non-Eclipse framework it may be considered as the
alternative bridge from Eclipse-based to non-Eclipse technologies at large.
IMHO, So far, the other approaches in Eclipse are not close to practical
solutions and I doubt that even go in usable by developers directions.

Guennadi Vanine
Previous Topic:creation review slides
Next Topic:Happy valentine's day!
Goto Forum:
  


Current Time: Thu Oct 23 14:46:06 GMT 2014

Powered by FUDForum. Page generated in 0.05513 seconds
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software