Sounds good. I'm not worried about different behavior, but just inconsistent implementation of that behavior internally
I think you are likely welcome to be a little aggressive in adding argument checks. If you flag a precondition that shouldn't be restricted, it is easy to discover and may well be something that should be restricted. On Fri, Aug 13, 2010 at 9:18 AM, Eugen Paraschiv <[email protected]> wrote: > Sure, makes sense to do this according to the boyscout principle and based > on patches. I will start working on such a patch for the code area I'm > working with, not for the whole project. As for an inconsistent state for > the condition checking logic, it should not be an issue, as the > Preconditions class throws the exact *same exceptions* as the ones that are > thrown manually, so in fact it should behave exactly the same for a client > of the class. > About clustering and classification, I have not worked with that portion of > the code yet - I'm focusing on recommendation algorithms for now. > Thanks for the feedback, I'll make sure to open the JIRA issue. > Eugen.
