I've made some steps towards being able to use ark-tweet-nlp easily as a library some time ago.
The first part of this was to provide a proper maven build that produces a self-contained jar (not custom zsh build script or hardcoded paths anymore; all dependencies are cleanly specified and I've made all required external files Resources that get bundled with the jar).
I got diverted onto other things before I could do part two (don't always output stuff to stdout etc.) but you might find the mavenization already useful in itself, so I decided to quickly merge in the last master and send you a pull request before my branch bitrots.
I'm happy to clean up things more if required, but first I wanted to see if there is interest in this patch.
Moving stuff around to idiomatic locations; killing unneeded libs
Moved stuff around some more to deal w/ jar dependencies that aren't …
Merge in default.
The adaptations to scripts are pretty hasty, but the scripts as are won't work
out of the box anyway.
thanks! definitely interested. haven't had a chance to look at it much yet... and I personally don't know much about maven-ization.
Hi, we've been using this patch, and noticed it's much slower than the old (stupidly simple) build script. Is there any way to make a mavenized build not be so slow? We're not maven experts...
Thanks for writing all this, it's very helpful. We're working on integrating in the maven support.
Hi, how does Maven know what's a local versus remote dependency? For example, looking at the pom.xml, there are comments that the edu.berkeley.nlp posTagger.jar is a "local" dependency. I was presuming that means it is supposed to copy from lib/. It seems to do so? How does it know it's a local dependency?
Hi, FYI, the 0.3 version (and current master) are mavenized, though there are still improvements that could be done.