Welcome aboard!
I've been wanting to test Sorcer against JGDMS to see if any
compatiblity issues pop up too.
Regards,
Peter.
On 27/06/2017 4:30 AM, Michael Sobolewski wrote:
Hi Peter,
I would recommend it strongly and would like to contribute to the project under
Gradle.
With Gradle automation it will have the great impact on future development with
many
new potential contributors that already use Gradle for large scale projects
based on Jini.
Regards,
Mike
On Jun 26, 2017, at 11:32 AM, Dennis Reedy<dennis.re...@gmail.com> wrote:
Hi Peter,
Congrats on all the work you've put into this project. Modularizing the project
is a big step forward. As you know I've been using Maven for my projects, but
lately I've found that Gradle provides a much more powerful, straight forward
and flexible approach for project automation, especially for multi-module
projects.
You can take a look at what a Gradle project would look like with River here
(https://github.com/dreedyman/apache-river-example). If you'd like I could work
with you and see what a Gradle version of JGDMS would look like, IMO it will
simplify the project greatly.
HTH
Regards
Dennis
Sent from my iPhone
On Jun 23, 2017, at 7:50 AM, Peter<j...@zeus.net.au> wrote:
This is what a Maven Build looks like:
https://travis-ci.org/pfirmstone/JGDMS/builds/246158857?utm_source=email&utm_medium=notification
All modules are also OSGi bundles, no split packages, no circular dependencies.
Yeah even phoenix is still there, no longer dependant on the Sun JVM
implementation, can run on any JVM now and uses JERI Endpoints by default.
The only remaining component that is Sun JVM implementation dependant is the
JERI Kerberos provider.
There's even a compatibility library for Jini 2.1, so people can upgrade and
migrate their code on their time schedule.
All the old ways of using Jini are still supported, such as classdepandjar,
preferred classloading, but now Maven and OSGi are much better supported too.
Oh yeah, security has been addressed, deserialization with input validation,
the latest TLSv1.2 cyphers, IPv6 Global discovery announcement etc.
Oh and anyone can build it now, with a simple one line argument. The build
also includes CVE security checks.
These are the features that were so hard to get acceptance for, but as it turns
out, you don't need to break backward compatibility in order to achieve it.
This is how I'd like River to be, of course if the community wants something
else, then I'll support whatever the community decides.
Regards,
Peter.
