I've been looking at your post on the Knime site I have two Knime nodes which can receive and publish data to AMP and functions to receive and publish data to Knime, written mainly for a POC but also to familiarise myself with writing extensions to AMP, it is fairly primitive and unfortunately I used Xtext to produce a DSL to describe any data manipulation that may be required before I realized that Xtext was about to change from ANTR to Rats as its parser. I will publish the code on Knime labs and AMP labs when XText stabilizes. This allows use of Knime/AMP functionality, and licencing to coexist without to much duplication, but with some performance cost
I've been looking at your post on the Knime site I have two Knime nodes which can receive and publish data to AMP and functions to receive and publish data to Knime, written mainly for a POC but also to familiarise myself with writing extensions to AMP, it is fairly primitive and unfortunately I used Xtext to produce a DSL to describe any data manipulation that may be required before I realized that Xtext was about to change from ANTR to Rats as its parser. I will publish the code on Knime labs and AMP labs when XText stabilizes. This allows use of Knime/AMP functionality, and licencing to coexist without to much duplication, but with some performance cost
Neat! Unfortunately the problem with KNIME with respect to GPL isn't just that there can't be direct API dependencies, but that they can't be provided in the same product. (And IANAL but my AFAICT from my reading KNIME isn't compliant period ) I guess GPL's position is that "though shall have no copy-left license before mine". It's a really bizarre kind of accidental politics. But that should at least mean that you can export data from one entirely separate application and import it into another.
As far as I know XText is still using ANTLR and the AMP dependencies are based on that version and will be through the next release cycle I would think. Also you shouldn't have to worry about what parser is being used as the code is all generated. I've been working on some DSLs for AMF as well for some time..need to get that released.
The comment was made by the xText project lead in his blog. My problem is I have two large grammars to convert (Matlab / Octave and SAS / R ) originally produced in Sablecc I don't like the prospect of two conversion tasks.
The comment was made by the xText project lead in his blog. My problem is I have two large grammars to convert (Matlab / Octave and SAS / R ) originally produced in Sablecc I don't like the prospect of two conversion tasks.
Wait, you're considering implementing grammers for Matlab and R? That would be enormously cool, but ambitious. I'm probably misunderstanding you..
No I don't think so, already implemented as a Knime node, using sablecc, it has proved popular with the mathematics / statistic community if you want to get a measure of its complexity a similar project exists on Sourceforge, JMatlab, it also uses Sablecc, I chose it to give Xtext a trial before using it in anger
No I don't think so, already implemented as a Knime node, using sablecc, it has proved popular with the mathematics / statistic community if you want to get a measure of its complexity a similar project exists on Sourceforge, JMatlab, it also uses Sablecc, I chose it to give Xtext a trial before using it in anger
Yeah that makes sense. Now, when can we get you as a contributor here?
The state of play here is that the development machine setup is due to be delivered at the end of this month, the modeling machines are in use by the interested parties who are producing a wish list. This is at the moment a Knime list, I expect this to change as the post grads join the modeling group next week. At this point in time I'm using the InterConnect to satisfy these needs and intend to concentrate my efforts on original requirements. We need to discuss the look and feel you expect to achive in AMP and the mechanisms you intend to put in place for development i.e. are we going to have a Knime like wizard to generate function skeletons, SDK, etc. Perhaps a new post should be started to discuss this.
We need to discuss the look and feel you expect to achive in AMP and the mechanisms you intend to put in place for development i.e. are we going to have a Knime like wizard to generate function skeletons, SDK, etc. Perhaps a new post should be started to discuss this.
Sounds good. Please feel free to open bugzillas as well. I've just spent a brutal week trying to get an IDE / SDK build for AMP finished. I've had SDKs before but we've had to put them together manually. This will be hosted at Metascape (non-Eclipse IP) and I'll post here when it's up.