As a quick note, some of the things that Brendan mentioned in the talk didn't end up making it in to Harmony. Here's a link to the meeting minutes:

https://mail.mozilla.org/pipermail/es-discuss/2011-May/01474...

Notably, arrow functions and paren-free blocks of code didn't make it in ... classes and comprehensions did.

Harmony is a state of mind ;-). Seriously, we were considering what to promote to "ES.next", very likely ES6. Arrows may yet make it, some on TC39 want to give them a late permission slip. If not, there's always ES.next.next if we can achieve consensus in TC39 on them.

Hey Jeremy, i know that there's been a lot of discussion on the es-discussion list that decided how that all shaped up, but i can barely keep up with the flood of mail that comes through. How do you manage it? Do you actually keep up with all the messages that come through?

Shamefully, I don't subscribe to es-discuss. I just visit the archive pages every once in a while, and read the threads that look interesting.

Hah, i'd actually just unsubscribed from es-discuss earlier this week. I couldn't handle the deluge. I'll consider doing something similar.

There is nothing inherently non-dynamic about classes. Python has classes and is fully dynamic.

Man, I'm sure I've watched Python conference videos at offline, downloaded from some button or link in the blip.tv web-page, but I sure can't find it now.

If you want to download the video without waiting for network buffering, or on a device without an Internet connection (or without a Flash plugin), the RSS feed has links to downloadable MPEG4 files:

http://blip.tv/jsconf/rss

Very inspiring stuff from Jeremy Ashkenas. Also, he's made talking-with-ones-hands into an art.

This just in: jashkenas is a great speaker too.

I'm not a huge fan of CoffeeScript in general, but one big plus in its favour would be the ability to write one set of CoffeeScript that compiles to JavaScript AND ES.next. That way you could serve the regular JS to older browsers, and the newer JS to browsers that support it.

It's an interesting idea, but how much of a benefit would there really be to serving ES.next code to people with modern browsers? I'd love to see benchmarks / realistic code size comparisons.