There’s An Open Source Software Census
Try saying that 16 times fast.
A new group of analysts is trying to assess how much, and what, open source software is installed. To participate in the census, go to OSSCensus.org and download their free software package, called OSS Discovery. Sounds like a submarine.
It’s written, for whatever reason, in JRuby, a Java implementation of Ruby (the standard, official Ruby implementation is in C), and is itself open source, although the Census Edition is different in some way from the regular edition. It runs on most popular platforms.
I didn’t audit the code myself, and the site doesn’t provide MD5 checksums or any of the other assurances that are typical when you’re trying to persuade people to download and run something. At least it doesn’t insist you run it as root.
Once you’ve unzipped the 50MB package, take a leap of faith and run it on your computer. Discovery scans the filesystem, comparing what it finds against its included library of fingerprints, to try to figure out what (if any) open source software you have installed. I’m not sure how accurate the fingerprint method is, or how thoroughly patched and customized a piece of software can be before it fails to be recognized as itself.
Discovery is not as easy to use as Debian and Ubuntu’s popularity-contest, which automatically (optionally) tracks and reports installed packages without scanning, but this approach makes more sense across a very heterogeneous set of installations.
Here is the user guide for the software, and here are the scantyish reported results so far. 231 machines have been scanned as of this blogging. Mine are not among them: I let the software run on my server for about two hours before I got tired of its RAM greed and killed it. I anticipate that compliance may be a problem for the Census.
See Also:
