Closed kennycason closed 7 years ago
@rzo1 definitely agree. As of late I have been using Eclipse Collections for everything, but will try and stick to only using Java libs as much as possible.
I worked on this in my forked repo (and did some heavy refactoring to extract some concepts in different maven modules). Maybe this is interesting and I can submit a pull request?
yeah sounds great. don't refactor the chinese tokenizer yet however, I'll pull that out next and move it into a different project kumo-tokenizers-chinese, or something.
Hi @rzo1 and @kennycason
Is there any progress in removing lambdaj
?
I am going to do it myself if you did not start it yet.
Thanks.
@howtimeflies-io I just added the pull request... did this 2 months ago... ;)
Thank you @rzo1
I did it myself just now 😄
Maybe it is also nice to update the maven dependencies (commons-io
, commons-lang3
and junit
) to their latest version in your pull request.
I think, I did an update on the dependencies in my other branch (extracting the heavy nlp dependencies) @howtimeflies-io
@rzo1
Cool! I will check that branch. Thanks.
Thanks guys! I just reviewed the PRs, some minor comments and I'll test and merge!
@kennycason If I can support updating documentation and/or changelog, just comment here with concrete tasks, so we can go for Maven Central soon :)
@rzo1 I'm planning on updating doc, testing, and deploying to maven central tomorrow afternoon. I am baby sitting some large hadoop jobs so have a bit of "free" during the afternoon. :)
@rzo1 Did you run any of the examples after you made this refactor? Mine are all stuck in infinite loops, or run out of heap space, and such. Logging is also not printing out anymore.
I'm going to debug through it and see what the issue is.
Maybe we can meet on IRC to discuss on demand finding? ;)
@rzo1 it's no problem. Given the lack of unit tests and size of refactor, I'm not surprised, nor concerned ha. Learnings so far:
Of course, by lack of unit tests, I mean, MY lack of unit tests :D
it's trying to create an array of size 12,468x4070 in CollisionRaster for each word, when this is supposed to just be the size of the canvas. Not sure why yet, but that certainly explains the heap space exceptions
I was expecting, that I may break something - can you tell me the name of the test-case? So I can start some investigations, too ? I am IRC available at freenode / #kumo
Please think about removing
lambdaj
as this drags some odd dependencies (i.e.hamcrest
) into the code base.