关于lucene发展和多语言实现的方向
多语言lucene的发展无疑是基于java lucene的。一切的功能特性和兼容性的问题都要以java lucene为主。java lucene是其他语言lucene发展的鼻祖。
那么多语言lucene的发展应该怎么办呢?
看看下面的文字吧:
There is a concerted effort to develop a SWIG Lucene and there is also
a CLucene and an active Lucene4C project. I was crazy enough to
contemplate a native Ruby port once upon a time, and developed some
low-level I/O code and then realized what a maintenance hassle it'd be
to keep up with the always evolving Java Lucene.
The PyLucene crew (credit where its due, Andi Vajda!) did something
quite amazing... using GCJ and SWIG to interface Java Lucene with
Python. This, in my opinion, is the way to future "ports" to any
language. Let Java Lucene be the base and all other ports derive from
it. I'm not sure why motivates Garrett with Lucene4C - and I certainly
do not want to discourage anyone from tackling this as at the very
least it is a great computer science exercise and surely a learning
experience for anyone attempting it.
If you continue with your port, you are going to have to face the
realization that you will always be behind the Java version in terms of
features and compatibility - unless you're able to implement the
features every time you see a commit message.
> For example, I know that portersteimmer is deprecated by snowball...
> exist
> other classes not worth to port now?
If you don't plan on spending every waking moment porting, why not join
forces with the SWIG Lucene folks and interface to that from Delphi?
> Something else I must know? The code is based in Lucene 1.4.3...
You're already behind and there has been dramatic changes with the
latest codebase that will be Lucene 1.9/2.0.
作者是:Erik , lucene in action 的作者。
作者的主要观点是:
1、最好利用类似PyLucene 的方式来实现lucene的多语言化。
2、Lucene 1.9/2.0 将会发生重大变化。(我正在翻译中),多语言的lucenene
要么很难在时间上和java lucene保持兼容,要么迁移到多语言的过程很辛苦。
每个commit,你都需要跟踪,然后修改......