Andrus Adamchik wrote:
I am back to the plugin engine topic again. Now it is out of the
hypothetical realm, as 3.0 development has already started.
First I wanted to address earlier suggestion by Ahmed [1] to use JPF
[2]. Cayenne *as a package* is released under Apache license. ASF has an
explicit policy that prohibits bundling (L)GPL software. We can't even
use Launch4j for the Windows distro as it has a binary component that is
distributed with Cayenne. The only "workaround" for the LGPL
"incompatibility" is to release the Modeler entirely outside of Apache,
which is certainly NOT what we want.
*****
Another candidate is Platonos [3]. Which is an excellent one, but as
Ahmed and myself pointed out, the project is dormant, as its developers
can't for personal reasons dedicate much time to it.
I contacted Platonos developers to get a better feeling of the situation
(I am cc'ing to Kevin and Evert who are the main force behind the
project). To be entirely accurate the project is not dormant - Evert is
developing a new version [4], which I still need to explore, and looking
into JSR 198 (A Standard Extension API for IDE, which itself seems
dormant :-)). And Kevin is consistently providing support on the user
mailing list. Essentially the situation is slightly better than I had
with JStaple some time ago (really passionate about the project, but not
realistic to do 100 things at once).
Yes, I (Evert) am working on the new version for time to time, just got
my bachelor degree and now I am a bit busy with job interviews. JSR 198
is designed for IDE's but a lot of things could also be applied to many
other kinds of applications so thats the reason I look a bit into it, to
see what could be useful for our pluginengine.
Kevin suggested donating Platonos code to Apache. In the ideal world
this would be a great idea. This is a very nice component that would fit
either in Jakarta commons or Cayenne (believe it or not, such commonly
useful thing as a plugin engine is nowhere to be found). Still with no
support from the original developers for the time-consuming incubation
process, I don't see this as an option at this time. Essentially this
would place the Platonos incubation burden on Cayenne committers. Don't
know about others, but for me this is out of the question at the moment.
While I'd be glad to mentor such project at the Incubator, anything
beyond that is unrealistic until Cayenne's own incubation is finished.
Maybe we can scout for volunteers on the Incubator list??
Anyways, IMO the best option would be to start with current Platonos
code, working out any issues with Kevin and Evert. Kevin also suggested
giving commit access to interested Cayenne committers. While I am not
psyched about taking over the project without any sense of direction on
my end (this is a bit un-Apache as well), it will be sort of a guarantee
for the ongoing support of CayenneModeler requirements. The first one of
those being Maven builds from repo that we could use in Cayenne.
Thoughts? Comments?
(Kevin, Evert, please feel free to chime in as well)
Andrus
Yes, I think what Andrus suggest is good idea, because of the
time-consuming incubation process as he told.
Looking forward to comments on our pluginengine and see where our
pluginengine can be useful for your modeling tool.
[1] http://objectstyle.org/cayenne/lists/cayenne-devel/2006/07/0032.html
[2] http://jpf.sourceforge.net/
[3] http://platonos.sourceforge.net
[4] http://platonos.cvs.sourceforge.net/platonos/ppe2/
