Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I'm doing a lot of python lately...but yesterday I compared one of my python lucene search programs with the java equivalent...and the Java one was considerably faster. Also, the fact that I could write the same program with Scala is enticing.

I feel like I am being drawn back to the JVM after pushing it away for so long. It really is an impressive system.



JVM not Java. Sounds right!


Doesn't PyLucene call into normal Java Lucene anyway? If so, why would you expect that the Python version would be faster?


I wouldn't expect it ever. Rapid development not rapid execution. However, I didn't expect the difference in speed to be so large, considering it is calling into the Java lucene.


This might be overkill, but would it be possible to setup a mini-Java Lucene process to handle the indexing and queries, and talk to your Python process over a socket?

Like I said, overkill, but it could be worth it.

I say this being in the same boat: coming from the JVM to Python and looking back at some of the things I'm missing.


How does an accelerated (e.g., psyco enabled) version of the python compare?

The Java one would probably still win, but an easy two lines may bring them closer to parity.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: