one more thing we can discuss are the close() methods of JCache impl. Did a first version but sometimes closing a JCache component will remove all items of the cache. Not sure it is intended in distributed mode.
Romain Manni-Bucau Twitter: @rmannibucau Blog: http://rmannibucau.wordpress.com/ LinkedIn: http://fr.linkedin.com/in/rmannibucau Github: https://github.com/rmannibucau 2014-05-02 17:17 GMT+02:00 Romain Manni-Bucau <rmannibu...@gmail.com>: > Hi > > globally a big +1 > > some more remarks inline > > > > 2014-05-02 17:13 GMT+02:00 Thomas Vandahl <t...@apache.org>: >> Hi folks, >> >> I took some time to look over the code of the JCache implementation and >> I have some suggestions for simplification (under the assumption that I >> understood the intention correctly). >> >> - JCS does not implement the same model of cache element expiry, however >> a few more existing features could be used. The question is how strict >> are the requirements of the JSR. I think that most of the functionality >> of JCSElement is already there in ICacheElement. > > > would be awesome to merge if you think it is possible. "strict" part > can be tested running TCKs (once the few missing assertNotNull and the > shutdown are in place - didnt check today). Only interceptor related > ones should fail. If ExpiryTest fails it is a regression. > >> >> - I'd like to move Statistics to the core. The cache statistics of JCS >> are too scattered and this class would very much improve this situation. > > Ok. Was expecting something to compute distributed statistics and not > only local ones. Do you think we can without breaking the whole > project? > >> >> - JCS has element event handling built-in. It works, however, for expiry >> and spool events only. I made a few modifications so that the additional >> events for create, update etc. can be added easily. > > great news! If we can move listeners to just be internal handler > listener adapters it would be awesome. > >> >> - The code produces quite a few warnings referring to type safety and >> possible NPEs. I guess this could be improved. > > go for it when you see it :) > >> >> - The code of the core and jcache modules compiles fine here and all >> tests pass (MacOS 10.6.8, JDK 1.6.0_65). The 1.6 compiler needed some >> additional support with CacheTest, however. >> >> - Wonder where the commons-math3 dependency comes from? > > shouldn't be needed IMHO > >> >> Bye, Thomas. >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >> For additional commands, e-mail: dev-h...@commons.apache.org >> --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
