Firefox 3.5 went to public release-candidate status earlier this week. But while the whole 3.5 branch was still under wraps, I was sticking my neck out and running the bleeding-edge nightly builds of the browser -- and was surprised at how un-beta it was.

My strategy was simple: Keep two copies of Firefox on hand at all times. One was the "Shiretoko" nightly-build edition, which updated itself regularly (at least once a day). The other was the public 3.0 branch of FF, which I kept as a reference copy and also as a backup in the event Shiretoko ever became wholly unusable. This never happened.

There were days when some beta builds were flakier than others, but most of that could be fixed with a cache cleaning. The single biggest instability wasn't even in browsing per se, but when typing in the address bar: occasionally, Shiretoko would crash when trying to retrieve a list of matches from the browser's history or bookmarks. But other than that things just worked, even on complicated sites like Facebook.

Maybe my outlook on the whole thing has been colored by my experiences with beta software in general. I'm used to beta-level software that crashes on launch, corrupts user data, behaves in ways that defy analysis or common sense (and sometimes good taste), so to use something that even in beta remained this consistently solid was downright thrilling.

Why would I do this for Firefox? Well, for one, it's one program I have open almost constantly and rely on for a broadening gamut of things. I'm still not a fan of the idea that the web browser can and will be a container for every kind of application I could use, but that doesn't change the fact of its current use or its manner.

But it also provides, if only provisionally, a peek sideways into the way Mozilla does software development. I wouldn't argue that running a beta is any substitute for being an actual programmer and inspecting source code, but it can give you some idea of how well-baked the product is even when it's still being baked. From what I've seen Mozilla's stuff even in beta is solid -- not anything you'd want to use in production (unless you enjoy living dangerously), but it's remarkable how good it is even in those stages.

If you think running an beta is -- or isn't -- a good way to get a quick-and-dirty grasp of a particular programming house's development style, let's hear from you.

InformationWeek Analytics has published an independent analysis of the current state of open source adoption. Download the report here (registration required).

Follow me and the rest of InformationWeek on Twitter.