For our application that meant bringing 1 of 3 application servers down and slowing things down in general. So we're stuck with 3.0.5.
From what I understand the Rails community has declared 1.8.x as a thing of the past (tbf it IS a thing of the past. But in a non-perfect world migrations our not always easy). This may be a good thing (it forced us to march towards 1.9.2) but I'm sure it has caused a lot of problems.
The commit mostly fixes that particular issue (and is included in 3.0.7), but as the article point out Rails 3 AR performance still needs quite a bit of work.
Judging by the comments on that commit and the comments in this HN post, I would say that many of the performance regressions are only happening under ruby 1.8.x.
By this time, I wonder what the rails developers should spend their time with: making the framework better for people running the current version of the language or fixing performance problems due to shortcomings of the old version?
If you can't upgrade to 1.9 and the performance of rails 3 isn't good enough for you to upgrade rails, why don't you just stay with 2.3 and upgrade once you are ready to move to ruby 1.9?
That way your unwillingness to upgrade ruby doesn't "waste" the time of the rails developers by causing them to fix issues that don't even apply to more and more people over time.
Note: i'm talking about regressions that only affect 1.8
I'm finding it's not that people can't upgrade to 1.9.2, it's that many don't want to.
I have a pretty decent-sized application running on REE and Rails 2.3.x. I've made my code fully 1.9.2 compatible. My specs take on average 3.5 times longer to run in 1.9.2. Profiling that shows that close to 60% of the time is spent in Kernel#require. Now, it's pretty common for people to say only profile in the production environment, but I need to be able to run my specs. And it's not acceptable to shell that out to a CI server or to run REE in dev. mode and deploy with 1.9.2. Running in spork is a gross approximation, too; I've seen way too many issues there.
My suspicion is that if you have a small enough project, the speed difference is negligible. If you have a large project, it becomes much more pronounced. So, it'd really be nice for me to continue to upgrade and improve my Rails-based app without having to drag along an environment that's going to hobble me.
FWIW, I do know people that have started on 1.9.2 and rolled back to REE because speed was such a problem.
I see. My two projects are indeed quite small yet (but running the tests already takes quite long).
But: the solution to this problem isn't not upgrading and then demanding support for your outdated environment. The solution is to fix whatever is slow and try to get a patch upstream.
Now I don't know how likely the ruby developers are to accept a patch, but once this becomes too painful for me, I will try and have a look. I might not be able to fix it, but at least I know that I'm not stuck in the past, terrified and unable to move
Well, the solution isn't upgrading for the sake of upgrading either. 1.9.2 buys me absolutely nothing and has some major costs associated with it. REE is chugging along like a champ.
I don't really understand why the release of 1.9.2 meant all else had to be dropped. Most other communities continue to support their stable releases. It's not as if 1.9.2 has even displaced 1.8.7 with virtually any of the linux distros either. If your policy is to use security-audited / supported packages, as is the case in many environments, moving to 1.9.2 is a dealbreaker.
Anyway, supporting 1.8.7 and 1.9.2 is trivial in most cases. I'm not demanding support for my environment, which I prefer to think of more as stable and battle-tested than "outdated." But I don't understand actively dropping support for it either.
For our application that meant bringing 1 of 3 application servers down and slowing things down in general. So we're stuck with 3.0.5.
From what I understand the Rails community has declared 1.8.x as a thing of the past (tbf it IS a thing of the past. But in a non-perfect world migrations our not always easy). This may be a good thing (it forced us to march towards 1.9.2) but I'm sure it has caused a lot of problems.