|Re: Triquetrum Components [message #1697926 is a reply to message #1697925]
||Tue, 09 June 2015 15:36
| Jay Billings
Registered: July 2011
From Christopher Brooks on the contribution from Ptolemy:|
In Ptolemy II, we have the notion of configurations. The Ptiny configuration is a small, useful configuration with a minimum of third party licenses. Ptiny includes the kernel, common polymorphic actors, common models of computation and Vergil, the UI. Vergil is AWT and is what we are looking at replacing with something that is more based on Eclipse.
The Ptolemy II source code is primarily BSD, so contributing it should not be that difficult. We've done a reasonably good job managing the licenses, the dependencies in the core are fairly straightforward. A tricky part could be tracking down certain contributors and getting an assignment of copyright. Fortunately, the core is fairly stable and not that many people have worked on it.
Hallvard Trætteberg created some OSGi modules that use the core Ptolemy II classes. One issue that we ran in to is that Ptolemy II uses the class path to add actors, thus we don't always know in advance the class names of all the actors that are to be used in a system. I believe that this is a bit contrary to how OSGi works, where OSGi expects to know all the dependencies in advance. There are workarounds to this though.
Jay Jay Billings
Oak Ridge National Laboratory
|Re: Triquetrum Components [message #1697946 is a reply to message #1697934]
||Tue, 09 June 2015 17:37
| Christopher Brooks
Registered: July 2009
We don't necessarily have to reuse the ptII code core code, though if we don't use it, then we end up reimplementing the wheel. There is quite a bit of engineering there, and some particularly subtle issues.|
Looking at the ptII code and using it as the basis for triquetrum would probably be a win. One big issue would be splitting things in to interface and implementation. It could be that in the new code, the package names, class names and method names might change from the names in ptII, but the bodies of a number of the methods would be similar.
It is ok if we lose backward compatibility between ptII and triquetrum. It would be nice if we had either a conversion script or had only a few types of changes to start with. Delaying the split for awhile could be a good thing.
In this message, I had a bunch of text with links to the triquetrum wiki that would help get the conversation rolling, but the forum says:
"You can only use links to eclipse.org sites while you have fewer than 5 messages."
Anyway, triquetrum.org links to a wiki that has some info.
Also, try searching for
Ptolemy Package Dependencies
the first link is a page on the kepler site that discusses the packages that are there.
BTW - I don't find forums that easy to work unless I get email about updates to the forum. I checked the 'Post Notification' button for this post, so we will see. If I'm not responding to a future post then it is because I probably because I have not seen it, so feel free to email me directly and I'll reply within the forum. Does this forum have an email gateway that would allow me to read and post?
Is there a wiki that we can use? As an alternative, I can update the package dependency work on the triq wiki and give out accounts to anyone who wants one.
Powered by FUDForum
. Page generated in 0.02409 seconds