Lambda the Ultimate

It doesn't seem to be an O'Reilly sponsored conference. Why is the event co-located with OSCON?

One question: How?

This is as close to "How" as I could see on their page.

Want to talk to your peers about the latest in JIT compilation techniques, type systems, source-to-source translation, or syntactic theory? Just propose a session and get your geek on with like-minded language gurus.

Hi, I'm one of the event's organizers.

We are indeed seeking to bring together language designers with VM authors. We've lined up Tav from the PyPy project, for exactly that reason. We'd love to see other VM folks there, and we're just as interested in the practicalities of creating new languages as the theoretical aspects of designing them.

Please do point any VM folks you know at our site. Thanks much!

My impression is that advancing the start of the art in programming language design and implementation, at least for hackers rather than researchers, is through low-level virtual machines like LLVM and VMKit

I think any attempt to quantify "advancing the state of the art" is self-limiting. LLVM and VMKit are branches on a larger tree. They are just one aspect of implementation, an aspect that many programmers are less interested in than the language itself.

Can you explain how implementation aspects of Parrot were divorced from Perl6 language specification?

At least here you can draw from real world examples you should be familiar with.

Otherwise, I fail to see what exactly you mean by self-limiting. The point behind being acquainted with tools like LLVM and VMKit, or knowing *Caml internals well enough, is to profit off decades of design knowledge and not reinvent the wheel. In doing so, you can spend months doing what used to take years. This makes labors of love more likely to produce something others can use, and not just watch the blood, sweat and tears drip.