On 11/19/2011 12:35, Orchid XP v8 wrote:
> It's not like you just write the code in
> Erlang and it magically becomes reliable.

No, of course not. You certainly have to write the code to be 
fault-tollerant. The difference is that in Erlang it's much easier, and you 
have bunches of libraries to help out.

> Well, if there's a module named "Cookies", then it's a pretty good bet what
> that does.

Yeah, if only everything was named that well.

> Yeah, that could work. I mean, it wouldn't work with my usual iterative
> prototyping method, but hey...

You would be surprised. You can still write a one or two sentence comment at 
the start of a function before you code it. It's no harder than deciding 
what to name the function and the arguments. And of course you go back and 
fix the comments as you code.

> result for every input you can think of, but it's rather harder to *prove*
> that it works for *all* inputs.)

Yeah. You can throw a bunch of random data at both, tho.

> OK. So before you erase the types, note down enough information to allow
> dynamic loading. It's not that hard. [People have managed to do it for
> Haskell, for example.]

Well, sure. As long as your interface is skinny enough that you can do that, 
then you're golden.

> Unfortunately, every PC on Earth has a web browser installed.

And, more importantly, every firewall has a hole for port 80. *That* was the 
real impetus.

> why Hotmail got started. (And as best as I can tell, that's where this whole
> crazy idea originated.)

Nah. There were lots of systems putting apps over http because it would go 
thru firewalls long before anyone tried to reimplement stuff that everyone 
already had software for like email.

> Ooo, and recording everything twice, for error recovery. (They call it
> "double-entry book keeping". We call it "write-ahead logging".)

Journaling. :-)

-- 
Darren New, San Diego CA, USA (PST)
   People tell me I am the counter-example.

Post a reply to this message