Wouldn't it have been easier to have moved away from PHP? From what I understand, Facebook only uses PHP for the most front of ends. Business logic is all in other languages. Isn't it easy to port the front-end into something else?
Facebook has a large, well-tested codebase, with a huge amount of infrastructure built on and to support PHP. It would be ridiculous to expect Facebook to migrate all of this to a different language.
Re: the article itself, you'll have to wait until Tuesday.
Twitter from Ruby into Scala and JVM: http://www.artima.com/scalazine/articles/twitter_on_scala.ht...
I think there are two aspects of a language (correct me if I am wrong) -- it's elegance/simplicity/breadth of code produced by programmers that write with it --- and then the efficiency/effectiveness/HW-optimized executable code produced by its compiler once it's run: so how does PHP and Facebook fit in with all this?
The only complex thing about Twitter is its size, and I bet their developers are working round the clock just to keep it from falling apart.
I've worked on many Rails apps, and have upgraded the apps from version to version. It's a pain when key elements of the API shift, but it's not that bad...even when the project has monkey-patched Rails a lot. And twitter certainly has the resources to afford to dedicate a few programmers to this task, so I'm just not sure I buy it.
Supposedly, the problem was caused by Cache Money, but nobody at Twitter wanted to risk moving to a different version again. They're still on 2.0 today. :-)
Another fun fact: Twitter has over 1,500 remote git branches. They also have bright green deer in the reception area of their office. :-)
Deep down inside I wish they were using ERB (or HAML) however, and he was writing HERB (or HHAML). I'm still enamored with Ruby syntax, and wish it would get some more big business love.