Machine VM + Cloud API - Rewriting The Cloud From Scratch | Hacker News Reader

Machine VM + Cloud API - Rewriting The Cloud From Scratch

highscalability.com

36 pointslosvedir9 comments

Given that there have been a few sprouts of Forth talk on HN lately, it's worth mentioning that an application-specific OS (written in an application-specific language) running on bare metal is sort of the quintessential Forth idea. Running it in the cloud, on virtual bare metal, would of course be Forth heresy, but still worth trying.

FWIW, every innovation in PL research coming out in the last 40 years can be considered anti-Forth, including binary security, strong-typing, platform independent code, modules, streams, exception handling and what have you.

I just went through Ertl's dissertation for the third time in as many years, and my impression is that Forth stands somewhere beside formal theories of programming languages, not within. His dissertation provides 3 implementations of forth, just as Dybvig provides 3 for scheme. The two papers are nearly identical in scope; showing different implementation techniques for a tiny kernel language. However, the Forth paper is entirely in C and assembly, without the slightest hint of formal semantics or any other mathematical reasoning. It's a hack upon glorious hack. You can't reason about Forth code without FULL abstract execution, to wit, implementing a full Forth VM with side-effects, or emitting 3-address-code, or some other linear tuple IR, and using traditional techniques from register-based languages with variable assignment.

Having said that, I can see why Forth might be fun, as a soul-cleansing language for a programmer who has spent too long in abstraction. Haskellers should use it to bring themselves down a notch [grumble, mumble mumble, something rude, under my breath]

[Edit:

Oh fuck, I forgot about Factor!

I pre-accept defeat, and retract my possibly under-informed arguments if and when Slava shows up to this discussion :-)

If his opinion differs, I don't think I am qualified to disagree with the sharpest dude hacking on a language implementation today, and Forth domain expert to boot]

This is an interesting concept. However, I think this article needs to define its target audience better. It seems to be speaking specifically to people doing web hosting (which you may or may not argue includes all of LAMP use-cases).

The idea that we can forgo operating systems because there are APIs to call doesn't make sense if your company heavily invests in infrastructure. Sometimes you need low-level access to hardware to fine-tune performance and scalability because your use-case is not the general use-case, even in a "cloud" environment.

Previous discussion: http://news.ycombinator.com/item?id=1762210

LtU discussion:

http://lambda-the-ultimate.org/node/4095

I posted a related thesis here a while ago: http://clubtroppo.com.au/2008/07/10/shared-hosting-is-doomed...

That the complexity of shared hosting would make it uneconomic vs VPSes in the long run.

Would be cool if it were javascript (it's OCaml), with nodejs I guess...

It targets javascript with Jake Donham's ocamljs.

The fact that it's not an everyday "job language" should really entice your inner "programming pervert"; hard to explain, but there is something erotic about seeing fringe languages in industrial applications. I would happily setup a SAP cluster if it had, say, a Forth repl hidden in one About box. I wasted countless hours in Excel, which I abhor, after it was mentioned in some numeric computing paper. And is there a better opportunity, on a boring weekend, than an oopsing Unix box with nothing critical on it :-)

Don't hesitate to delight yourself, no one will do it for you.

Reply on news.ycombinator.com