Re: DocSearcher on Debian
On Tue 17 June 2003 04:50, Bruce Best (CRO) wrote:
> We are currently in the process of revising the site, and moving from a
> primarily static HTML site (which, with approximately 30,000 documents is
> becoming a bit hard to manage in FrontPage) to a PHP/MySQL site using
> Lucene for the search function. I was interested to see DocSearcher (which
> uses Lucene) included in the suggested programs at Debian-Lex, and decided
> to give it a test on my Woody system.
> I started with trying to create a searchable CD of some PDF files, which
> seemed to be the most straightforward way to test the program. I have
> Blackdown j2re1.3 installed, and was able to run the program with "java
> -jar DocSearcher.jar", which created an index in /home/bruce/.docsearcher,
> and launched the GUI for DocSearcher
> However, when I tried to run a search for the term "constitutional", I got
> no results. I got the following error in the terminal I started DocSearcher
> STATUS:SEARCH STRING:+body:(constitutional)...
> at DocSearch.doSearch(DocSearch.java:1727)
> at DocSearch.handleEventCommand(DocSearch.java:1002)
> at GuiThread.run(GuiThread.java:29)
> at java.lang.Thread.run(Thread.java:484)
> Knowing almost nothing about Java, I'm not sure what this error means.
> Anyone managed to get DocSearcher running on Debian?
I'm using Debian sid, up to date a couple of weeks ago. I've got Blackdown
j2re 1.4.1-beta. When DocSearcher searches on "Title and Document Body" it
fails to produce results (even though it should), although there is no error
message. Searches on "Document Body" alone succeed.
I must admit that I haven't used DocSearcher much, but so far it seems to be
stable and a worthwhile addition to Debian/Debian-Lex. I'm not entirely
clear on where java in Debian is up to. Quoting from Drew Scott Daniels
Sarge release issues message:
"3 Java (some packages were in contrib/non-free because they wouldn't work
with free Java implementations (ie kaffe). kaffe is free but marked
[REMOVE], (Sun's jre isn't?), gjc and gij need some work acording to
An old, but still relevant java faq is at
http://www.debian.org/doc/manuals/debian-java-faq Is Kaffe waiting for a
DSA before it's RC bugs are fixed?"
So, I assume if DocSearcher is to be packaged for Debian it will be in
non-free until something like Kaffe is ready. Jeremy, do you know anything
about the java situation? Maybe we need to test DocSearcher with kaffe, gjc
& gij. I have zero experience with any of those, but if/when I get some
time, I'll give it a try.