I don't think the solution regarding segfaults on a search of only
stopwords is the right way to solve the problem. The problem is in
clucene or an inappropriate use of it. A bug should be entered into the
clucene project for it.
I found a related, and now fixed bug, where "a AND b" where "a" is a
stopword would produce a segfault. See:
http://sourceforge.net/tracker/index.php?func=detail&aid=1051984&group_id=80013&atid=558446
I'm not saying that removing stopwords shouldn't be done. It should, but
it is a different issue.
Presuming maintaining bc on existing indexes is a good thing, it needs
to be fixed in clucene.
BTW, BibleTime is using 0.9.2 which is akin to Lucene 2.3.x. Upgrading
to what is getting the most developer attention might be a good thing.
In Java Lucene, they guarantee that a 2.x reader can read a 1.x built
index and the StandardAnalyzer is also backward compatible in these
series. I'm nearly sure the same is true with clucene.
I found in going from Lucene 1.4.x to 2.x that there were very minimal
changes to be made. I expect that given our trivial use of clucene that
it should be trivial to upgrade to 0.9.x. The BT folks should be able to
give an idea as to how stable it is.
In Him,
DM
On 08/30/2009 11:17 PM, Jonathan Marsden wrote:
I have split out three specific patches from Matthew's single larger
patch, and tested what happens when I incorporate them in our sword
packages. They all test out fine for me. Each one fixes a clearly
present bug in the current packaged code.
The three patch files are available for download at:
http://crosswire.org/~jmarsden/16_larger_search_buffer.diff
http://crosswire.org/~jmarsden/17_no_stop_words.diff
http://crosswire.org/~jmarsden/18_check_index_dir_is_writeable.diff
Although they were made against the sword 1.6.0 tarball code by quilt,
they apply cleanly against SWORD svn head (use patch -p1<filename.diff
to apply them by hand). I'd be happy to see them accepted into the
official SWORD codebase.
I did this work because I'm seriously thinking about trying for an FFe,
so we can get these, plus a packaging change to re-enable module
encryption and decryption (!), into Karmic.
Anyone with objections, or anyone with suggested additional bugfixes
that should go into such a package update, please speak up.
Jonathan
_______________________________________________
sword-devel mailing list: sword-devel@crosswire.org
http://www.crosswire.org/mailman/listinfo/sword-devel
Instructions to unsubscribe/change your settings at above page
_______________________________________________
sword-devel mailing list: sword-devel@crosswire.org
http://www.crosswire.org/mailman/listinfo/sword-devel
Instructions to unsubscribe/change your settings at above page
