On October 1, 1999 at 14:15, "Simeon ben Nevel" wrote:
Since I have kept (off-site) virtually every messages since the list
began, I plan on building an archive on my Linux box and burning the whole
thing to a CD-Rom.
The problem is that I wouldn't be able use use a regular cgi-bin based
search engine (like Wilma/Glimpse or MarcSearch) against the archives on
The "cgi-bin" aspect is not the real issue since such programs
can be modified to run in non-cgi environments. I am guessing the
real problem is having a platform-neutral search engine. I
am assuming the CD-ROM can be used on variety of platforms. Or
will you cater to just a particular OS?
I have since realized that a Java based applet would fill the bill.
Since I believe you want a search engine to function within
Web client (hence the idea of using Java), you could have a small
web server on the CD that the persons starts for search-based
Something like HTTPi (http://stockholm.ptloma.edu/httpi/) could
be used. It is a pure Perl webserver. I think currently, it
only functions on Unix-type systems since it uses fork(). Unsure
if Perl 5.6 will allow fork()-based programs to run (effectively)
An advantage of the mini-server approach is that clients that
do not support Java (like Lynx), or Java is not activated for
security concerns, can do searching against the archives. No
client-side applet is needed. Regular HTML forms can be used.
BTW, a Java applet will not be enough. Java puts many restrictions
on applets for security reasons, so you will not be able to due
file I/O. You will need server the web client can communicate
with, where the server does the file I/O operations.
Unhappily, an extensive search has shown that the available
(commercial/share/free-ware) engines either don't serve my needs or are
prohibitively expensive (especially as I might be providing copies of the
archive to 100's of people).
With how Java has been deployed, you have an atmosphere where many
Java programs are proprietary. Not as open source (free) as the
In looking around for alternatives I ran across a number of sites where
one could "request" some software for open development, solicit
"proposals" from developers and gather "interest" from other folks who'd
like to see the software built. Interested parties commit to putting up
some bucks to fund the development.
I figure if we could get 20 or 30 people to put up $20.00 (USD) each, we
could probably attract a Java developer to do the work.
If you'd be interested in helping support(fund) and/or develop
specifications for a MHonArc-specific Java based search engine please drop
me a note. If there is sufficient interest, I'll set up a mailing list at
one of the freebie hosting services where we can hash things out.
If you take the search index approach, you will need a
platform-independent index file (if you are supporting multiple OS's).
If an existing search engine can create platform-independent index
files, all you would need to build is the search client.
If you think I'm entirely out in left field, forgive the intrusion, but
please *do* drop me a note and tell me why you think this wouldn't work.
O'reilly uses a Java server application for the The Perl CD Bookshelf.
I did a search an came up with Jetty,
<http://www.mortbay.com/software/Jetty.html>. It is Open Source. I
guess as log a system has the applicable version of JDK, or the Java
VM, it will run. This could be the base for the server component, and
then all that is needed is the software to handle the searching.
P.S. I am taking a Java class next week at where I work. It
looks like I can no longer avoid it. Hopefully, the class will