Home About Eric Topics SourceGear

2006-10-11 14:15:59

Google Code Search

Google made a nice splash last week by announcing Google Code Search, which allows searching public source code.  Cool.

Google is big and powerful, so everything they do gets lots of press.  But to be fair, it should be noted that Google wasn't first in this market.  Several other players were up and running before Google's offering:

In fact, from my own highly biased perspective, none of the above was first to market.  As far as I know, the first source code search engine was CodeCatalog.

Sadly, if you just clicked on the link to visit CodeCatalog.com, you found a squatter page.  CodeCatalog is now gone.  In fact, its life was rather short, but it predated all the current players by five years or so.

CodeCatalog was a pet project of mine back in late 1999 and early 2000.  I got rather obsessed with the idea of source code search engine.  I'm sure I was supposed to be acting more like a CEO of SourceGear, but instead I was hacking on parsers and indexers.  CodeCatalog was like my personal skunk works project.  I designed and wrote the code.  I coordinated the marketing splash.  I was the sales guy who closed the deal.  The whole thing was a reckless risk, but it turned out very well for SourceGear.

The front end for CodeCatalog was built with Zope.  IIRC, the backend was written in Java.

Performance issues were tricky, and I was only indexing around 20 million lines of code.  The current players are handling ten times that amount or more.  Nonetheless, most of my development time was spent making sure that searches were blazingly fast.

I never really intended to run a source code search engine as a business.  My goal was to unveil CodeCatalog with a gaudy launch at LinuxWorld Expo in February 2000 at the Javits convention center in New York.  All the big Linux and Open Source players would be there.  We would draw lots of attention to CodeCatalog and then sell it to somebody like SourceForge for big bucks.

We decided to build the brand identity for CodeCatalog around a cartoon mascot.  We hired a cartoonist to draw a robotic dog.  We named him Cody.

At LinuxWorld, Cody was everywhere.  We did big splashy booth graphics and printed thousands of T-shirts.

And we even brought Cody himself.  Sort of.  The role of Cody was played by a Sony Aibo.  We gave the dog away as a raffle prize one the last day of the show.  Back then, Aibos were relatively new and difficult to find.  I had to buy one on eBay.

As it turns out, things worked out very much like we planned, but not exactly.

Some time before LinuxWorld Expo, we crossed paths with a company called OpenAvenue.  Long story short, we closed a deal to sell CodeCatalog to OpenAvenue just before the LinuxWorld show.  I should not disclose the sale price, but I will say that it was a big enough bag of cash that we have no regrets.

Unfortunately, OpenAvenue languished.  Their business was very roughly in the same space as SourceForge and CollabNet, but they just didn't get enough traction.  Ultimately, OpenAvenue went out of business and took CodeCatalog down with them.

Fortunately for us, OpenAvenue met their financial obligations to SourceGear even as their own company was going through a very difficult financial struggle.  I was very grateful to their CEO Paul Hessinger for that.

Paul and I have not kept in touch, but I just did a Google search and discovered that he is currently CEO of InRule (they sell a rule engine for .NET).  Their management team page also lists Kate McGowan, who I remember as the primary marketing person for OpenAvenue.

A search of the Internet Archive Wayback Machine shows that the earliest archived copy of the CodeCatalog website is from May 2000.  The last version was from March 2001.

If Cody were still alive today, he would be almost six years old.  If the technology and website were actively maintained and improved throughout that time period, CodeCatalog would probably be quite impressive.  I'm happy with the deal, but I do regret that Cody never reached his potential.

But business realities are what they are, and CodeCatalog is long forgotten by almost everyone.

Except me.  I still have several of the T-shirts, and I sometimes wear one when I'm just hanging around at home.  I guess it's my way of keeping Cody's memory alive.  :-)