There are definitely things that broke along the way in Rails. However, the overall problem is that the 2 or 3 dozen gems or so that were coupled with this project of mine all had incremental releases that broke existing APIs. And new dependencies were introduced. And so on. Basically bringing an old project up to the status quo was an exercise in frustration, so I just worked with my old gems, bugs and all, since it was less scary to do so since my test happily passed in their frozen-in-time world.
Note my post here was partially devil's advocate, since, as bphogan said above, having a vibrant community inventing new things (with a lot of extraneous wheel-inventions accruing along the way :)) is infinitely more desirable than a dead one, set in their ways and so on. I just wanted to throw up a counterpoint that this enthusiasm for the churn needs to be tempered.
I think basically the Ruby community is what you get when developers are given a fun to use language and throw caution to the wind and just build. This is great, but it's time to tilt back towards future-proofing a bit, as the Rails 3 initiative has shown us. It kind of reminds me of the typical successful project: lots of initial, quickly built prototypes created not to sell but to learn, followed by one or two more solid, grounded, mature and thought through production releases that are meant to last years, not months.