[FoRK] the fix is in? <rant> lang, os, etc. continued </rant>
dmorton at bitfurnace.com
Thu Dec 10 20:07:15 PST 2009
amongst the first to be put up against the wall and shot by future
On Fri, Dec 11, 2009 at 3:49 AM, Jeff Bone <jbone at place.org> wrote:
> Okay, so here's the way I see it.
> with e.g. generators / etc., commonJS, narwhal, node.js, er.js, competitive
> work in the engine space (particularly v8) etc. --- all headed in the same
> direction --- not converging yet, different wrinkles in each. But the
> trends seem clear:
> - js is leaving the browser behind
> - js is gearing up to become a general "systems programming language"
> - js solves the adoption problem implicitly
> - ubiquity - some variant runs everywhere
> - familiarity - everybody's written a little bit
> - *good enough* --- for just about anything?
> - big dogs pouring resources on it
> - but *can* it be made to be performant enough?
> - most performance is architecture, not code paths
> - can it be made to play nicely w/ native-code for fast paths?
> - but of course: it always *has* done exactly that!
> We're moving a meta-level up the technology stack, same sort of move as
> assembler-to-C. With appropriate new engine technologies like V8, it's
> looking like js could be used for all but the very-close-to-metal apps.
> Hell, node.js isn't even to version 0.2 and it already gives thin a run for
> its money; I like where this is going.
> possible for years now, just like many of us. It's got a lot of cons, not
> the least of which is bulk (node is ~2.5M by default on OSX, ugh; compare /
> contrast w/ tiny Scheme VMs ala Wasp and Hedgehog.) But: perhaps it's time
> to acknowledge the writing on the wall...
> I'm not sure how I feel about all this, but I think it is what it is... I'd
> like to see js evolve in the right direction --- tail recursion optimization
> a required feature, first-class continuation support, reified interpreter /
> vm / program state, cross-engine standardized FFI, etc. I'm not sure it
> will get there. (I'm not sure it has to, but...) Those would provide a
> better substrate language for future systems.
> But maybe we don't need all that fancy stuff, maybe it's good enough as-is.
> I see a sort of stack evolving that looks something like this:
> - host OS
> - bare-metal
> (Nb., the above looks a hell of a lot like the Inferno stack... ala:
> - interactive user shell (Inferno sh, in Limbo)
> - Limbo + runtime Inferno "user OS" / environment
> - Dis VM (eventually bootstrapped)
> - host OS
> - bare-metal
> Rather than continue down the road of dithering on substrate and vacillating
> between different toy implementations on different substrates, maybe I
> should just byte the bullet.
> Thoughts? Is js the "next" *real* "systems programming" language?
> PS - for the progeddit trolls, should this get cross-posted and should you
> stumble upon this... enjoy. Please note the rant tags. Take this w/ a
> grain of salt, and in context please. Just musing out loud...
> FoRK mailing list
More information about the FoRK