site-packages/lucene/*.jar files

[Jacob Floyd <te...@gmail.com>]

Hello,
Now it's time for the next saga in my gentoo pylucene ebuild.

Once I've compiled pylucene on gentoo (it works at last, hooray), it
installs the following files:
/usr
/usr/share
/usr/share/pylucene-2.4
/usr/share/pylucene-2.4/lib
/usr/share/pylucene-2.4/lib/extensions.jar
/usr/share/pylucene-2.4/package.env
/usr/lib
/usr/lib/python2.5
/usr/lib/python2.5/site-packages
/usr/lib/python2.5/site-packages/lucene
/usr/lib/python2.5/site-packages/lucene/_lucene.so
/usr/lib/python2.5/site-packages/lucene/__init__.py
/usr/lib/python2.5/site-packages/lucene/lucene-core.jar
/usr/lib/python2.5/site-packages/lucene/lucene-analyzers.jar
/usr/lib/python2.5/site-packages/lucene/lucene-snowball.jar
/usr/lib/python2.5/site-packages/lucene/lucene-highlighter.jar
/usr/lib/python2.5/site-packages/lucene/lucene-regex.jar
/usr/lib/python2.5/site-packages/lucene/lucene-queries.jar
/usr/lib/python2.5/site-packages/lucene/lucene-instantiated.jar
/usr/lib/python2.5/site-packages/lucene/extensions.jar
/usr/lib/python2.5/site-packages/lucene/__init__.pyc
/usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info
/usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/PKG-INFO
/usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/not-zip-safe
/usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/top_level.txt
/usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/dependency_links.txt
/usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/SOURCES.txt

That's great, it works well. However, those jar files are just copies
of the originals right? (diff says so). Is there any mechanism in
place already to place a symlink to the original jar files instead of
a copy? Would that have any negative impact on things?

I ask, because gentoo frowns on copies of jars floating around,
possibly getting out of sync one with another.

Thanks,
Jacob Floyd

Re: site-packages/lucene/*.jar files

[Andi Vajda <va...@apache.org>]

On Feb 16, 2009, at 12:26, Jacob Floyd <te...@gmail.com> wrote:

> Hello,
> Now it's time for the next saga in my gentoo pylucene ebuild.
>
> Once I've compiled pylucene on gentoo (it works at last, hooray), it
> installs the following files:
> /usr
> /usr/share
> /usr/share/pylucene-2.4
> /usr/share/pylucene-2.4/lib
> /usr/share/pylucene-2.4/lib/extensions.jar
> /usr/share/pylucene-2.4/package.env
> /usr/lib
> /usr/lib/python2.5
> /usr/lib/python2.5/site-packages
> /usr/lib/python2.5/site-packages/lucene
> /usr/lib/python2.5/site-packages/lucene/_lucene.so
> /usr/lib/python2.5/site-packages/lucene/__init__.py
> /usr/lib/python2.5/site-packages/lucene/lucene-core.jar
> /usr/lib/python2.5/site-packages/lucene/lucene-analyzers.jar
> /usr/lib/python2.5/site-packages/lucene/lucene-snowball.jar
> /usr/lib/python2.5/site-packages/lucene/lucene-highlighter.jar
> /usr/lib/python2.5/site-packages/lucene/lucene-regex.jar
> /usr/lib/python2.5/site-packages/lucene/lucene-queries.jar
> /usr/lib/python2.5/site-packages/lucene/lucene-instantiated.jar
> /usr/lib/python2.5/site-packages/lucene/extensions.jar
> /usr/lib/python2.5/site-packages/lucene/__init__.pyc
> /usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info
> /usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/PKG-INFO
> /usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/not-zip- 
> safe
> /usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/ 
> top_level.txt
> /usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/ 
> dependency_links.txt
> /usr/lib/python2.5/site-packages/lucene-2.4.0-py2.5.egg-info/ 
> SOURCES.txt
>
> That's great, it works well. However, those jar files are just copies
> of the originals right? (diff says so). Is there any mechanism in
> place already to place a symlink to the original jar files instead of
> a copy? Would that have any negative impact on things?
>
> I ask, because gentoo frowns on copies of jars floating around,
> possibly getting out of sync one with another.

If you don't want to include the jars, you don't have to. The trick,  
then, is to get the Python user to maintain a valid Java CLASSPATH  
that helps with finding them. This gets hairy quickly, is error prone,  
and is something a Python programmer doesn't - understandably - want  
to do.

By including them in the egg, you're getting two things:
   - lucene.CLASSPATH is correct
   - no getting out of sync with PyLucene since the jars it was built  
with are included in it.

Andi..

>
>
> Thanks,
> Jacob Floyd