Hi Rafa and Karl, To give more flexibility, It make sense to externalize models and go with configuration information. I'll modify the connector, so that users can provide models.
Thanks, Chalitha On Fri, Dec 18, 2015 at 5:30 PM, Karl Wright <[email protected]> wrote: > If the models are immutable resources, then we should probably just bundle > them in the jar as resources. If they are modifiable and/or it is expected > that the user will supply their own, then configuration information is the > way to go. > > Karl > > > On Fri, Dec 18, 2015 at 6:58 AM, Karl Wright <[email protected]> wrote: > > > Hi Rafa, > > > > I'm also not really happy with how the connector locates the models. > It's > > not going to work very well in a distributed environment. It seems like > > there's a missing piece here: the models should probably be imported as > > part of the connection's configuration information, rather than just an > > implicit file reference. How big are these model files, typically? > > > > Karl > > > > > > On Fri, Dec 18, 2015 at 6:53 AM, Karl Wright <[email protected]> wrote: > > > >> Yes, please create a ticket. But I'd create a branch and import the > code > >> into that first. I have to get licenses and build process working > before > >> we can bring everything into trunk. > >> > >> Thanks, > >> Karl > >> > >> > >> On Fri, Dec 18, 2015 at 6:38 AM, Rafa Haro <[email protected]> wrote: > >> > >>> Hi all, > >>> > >>> I was about to open a Jira ticket for including Chalitha's OpenNLP > >>> Transformation Connector ( > >>> https://github.com/ChalithaUdara/OpenNLP-Manifold-Connector) into the > >>> trunk. > >>> > >>> Everyone agree? > >>> > >>> Cheers, > >>> Rafa > >>> > >> > >> > > > -- J.M Chalitha Udara Perera *Department of Computer Science and Engineering,* *University of Moratuwa,* *Sri Lanka*
