Hacker News Reader

Comment by cookiecaper | Hacker News Reader

>We made one additional award – to PyPy, the Python JIT compiler, for $200,000.

This is great. I sincerely hope the PyPy developers see the light soon and start focusing on Python 3.x support, and that this money can help get it done. Haven't written a Python 2 app for something like 4 years? Python 3 is mature now.

I also sincerely hope that this portends Firefox bundling a Python runtime with the browser, providing a DOM API, and letting sites provide client-side scripts in Python. JavaScript has been allowed to reign for much too long.

View on HN

> This is great. I sincerely hope the PyPy developers see the light soon and start focusing on Python 3.x support

I don't think the lack of Python 3 support was reflective of a value judgment about Python 3, so much as that most of the companies willing to fund PyPy development so far are still on Python 2, and were more interested in funding other kinds of work (performance, numpy, cpython extension compatibility, etc.)

> I also sincerely hope that this portends Firefox bundling a Python runtime with the browser, providing a DOM API, and letting sites provide client-side scripts in Python. JavaScript has been allowed to reign for much too long.

This strikes me as extremely unlikely. There's zero appetite for introducing non-standard language runtimes into browsers these days as far as I can tell. But I guess you never know.

It's a bit rude to say there is zero appetite for something in reply to a person who just expressed an appetite for it.

I read it as "There's zero appetite [on the part of browser vendors] for introducing non-standard language runtimes into browsers these days".

Anyway, WebAssembly [0][1] is coming through the pipeline with the goal of opening a common browser runtime to more languages. Instead of Mozilla incorporating a Python runtime into Firefox, a Python runtime will eventually be able to target WebAssembly. For Python specifically, the work in progess on builtin GC support should help a lot [2].

As for the donation to PyPy, Mozilla uses Python heavily in internal tooling so I imagine that's part of the motivation.

[0] https://webassembly.github.io/

[1] https://bugzilla.mozilla.org/show_bug.cgi?id=1188259

[2] https://github.com/WebAssembly/design/blob/master/GC.md

Will WebAssembly allow for self-modifying code? Otherwise you're stuck with CPython, apparently.

The python runtime is fairly heavyweight, on the order of megabytes, unless I'm mistaken. This would be something that would be downloaded with every page request (although it may be cached in most cases)...

So I'm not sure it will be something you will want to use in a lot of places.

Not going to happen. Web Assembly is the one "alternate runtime" (which isn't really an alternate runtime) that is ever going to exist.

'twould be interesting if one could take a JIT'd or even partially JIT'd python program as input and get WASM out. You could have it be like a http server feature.

Someone developed asm.js backend for PyPy, so it's definitely possible.

There is nothing stopping anyone from making python emit wasm.

"I also sincerely hope that this portends Firefox bundling a Python runtime with the browser, providing a DOM API, and letting sites provide client-side scripts in Python."

I expect WebASM to eventually get DOM bindings. At that point you'll compile your Python into WebASM.

Yes, this is indeed on the roadmap:

https://github.com/WebAssembly/design/blob/master/GC.md

Interoperability with GC systems is a prerequisite and will come down the road from the initial ("MVP") release.

Google tried this with Dart and ended up backing out. That was the only real chance at having a commercial browser include a language runtime other than JS. WebAssembly is where it's at.

Reply on news.ycombinator.com