[JDEV] [jadmin] The Road to JabberSearch (BETA!)

[Justin Mecham]

Here I am, after quite possibly 4 months of "I'm really going to get this thing done.  Serious!  For sure next week!  Really!"  I've decided that the best motivation for getting JabberSearch completed and rock solid, is actual user input and complaints that this and that feature doesn't work like it should! =]

Let me back up here, back in January of this year at the last Jabber developers meeting up at Ryan Eatmon's place in Dallas, I began working on a PHP wrapper for the SWISH-E search engine that Jer had just started setting up.  I got it working before I left, and even set it up on Jabber.org, but found it to be very weak in comparison to other engines.  About this time I started thinking about starting a standalone site with a single index that would power the search for all the Jabber sites, and allow for searches to be performed on the same index from any of the sites or from JabberSearch itself.  I began to search for some other search engines and came up with mnoGoSearch after seeing it over at MySQL.com.

I installed mnoGoSearch and played around with it for a little while, but I felt limited by it as well.  I did liked the fact that it stored its index in a database, and that it had built-in PHP functionality, but these features alone didn't quite compare the the general weakness of the search engine itself.  I'm looking for Google in a box here, and nothing's coming close!

Upon more research I remember ht://dig. It's a great little search engine, a bit old and rusty, but it functions.  After I set it up I was pleased with the results, especially the page excerpts for matching documents, and decided it would hold up to the task of creating JabberSearch.  So I spent about 3 days implementing it, writing some custom templates that I could parse using PHP, then gaining some sense and changing the templates to XML.  I put together a rudimentary XML parser and everything was working great, well, until my idle-normally-reading-the-latest-news-on-the-web time brought me to our next search engine!  I never thought I'd get this much experience with search engines software, nor was I ever planning on making this much work for myself...

Enter ASPSeek...  Wow... I can't possibly say enough good things about this search engine.  It's the apex of user-installable search engines, the Jabber of functionality, and the Linux of price and performance.  It's my Google-in-a-box.  I'm still amazed at how wonderful it is..... Oh I'm sorry, here is a towel to wipe up my drool, I hope I didn't damage your keyboard.  Anyways, so I set up ASPSeek, was very impressed, found some bugs, requested some features, and get this... got some actual interaction with the developers!  Wow!  I like!  If you need a search engine, there is nothing better than ASPSeek: http://www.aspseek.org/  (I'm trying to get them to implement native XML output capabilities...)

So, the definitive-all-comprehensive-and-hopefully-widely-used-but-not-too-bandwidth-hogging-since-it's-currently-on-a-limited-bandwidth-box (gasp) JabberSearch has reached beta status!

JabberSearch is currently indexing just over 15,000 pages across 6 web sites.  Both the site frontend and backend require a lot of work yet, but the search facilities are functional.  My decision to announce is based on the fact that the recent Jabber.com survey and numerous mailing list posts demand a good search engine.  Like most of the cool software currently out there, this is work in progress and may break at any time.

For the next week or so JabberSearch will be in beta testing.  Once I'm confident that all the showstoppers are ironed out and the site design is finished then it'll go live for real.  Once this happens, JabberCentral, Jabber.org, Jabber DevZone, the Jabber Docs Site, the Jabber Mailing List Archives, and a couple other sites will be powered by JabberSearch directly.  In the future, the database will be open for anyone to use and query locally on their own sites or applications, by way of XML-formatted search results.  I've got some cool ideas cooking in my head, but first things first.

So pull up a web browser and walk, don't run, over to http://www.jabbersearch.org/ and find those bugs! When (I say this because I'm sure you will) you find bugs, or if you have any comments or [de]constructive criticism, you can e-mail me or jabber me at justin at jabber.org.  At least I think you can e-mail me there, I've never tried it really.  Just reply to this one to be sure :)

Go easy, and be sure to heed to the practitioner's warning! Enjoy!

Justin Mecham
JabberCentral Network Manager
http://www.jabbercentral.org/


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.jabber.org/jdev/attachments/20010505/274432e6/attachment-0005.htm>