On the other hand, a lot of the stuff at Google is Java. Also, I can't imagine writing a reliable 1m lines of code system (say for a bank) in Lisp (or any other dynamic language for that matter).
They also do some Common Lisp on the airline search division.
btw, Common Lisp is a "strong typed" language, and many large enough code-bases were here. Lisp machine ecosystem is a good example.
A lot of banking software has been written in 4GL and stored procedures (SQL). When Oracle stuffed its products with Java it was not necessarily an improvement, rather things became more difficult and messy than before.
I'm not sure how widely Go is being chosen (over say Java) for new projects inside Google. Maybe some Googler could clue us in?
> btw, Common Lisp is a "strong typed" language,
Can you point me to a code example? The ones I did find didn't have any type information in them.
> A lot of banking software has been written in 4GL and stored procedures (SQL). When Oracle stuffed its products with Java it was not necessarily an improvement, rather things became more difficult and messy than before.
Banking was just an example, and I'm sure they are plenty horror stories there involving Java (esp. when used in some awkward way).
What I wanted to say is that IMO the effort to build and maintain large code base in for example Python would typically be much greater than in Java. For small code bases, I'm not sure who wins - the development phase will obviously be faster in Python (which is the only thing that matters for startups - hence python/ruby/... popularity here), but maintenance will still be a pain - after the initial devs leave, hires will need to learn that Python codebase, and lack of type information will make it harder.