Re: Lua and Saint-Exupery?

[Jamie Webb <j@...>]

On Fri, Feb 25, 2005 at 10:56:09PM +0000, David Given wrote:
> On Friday 25 February 2005 21:23, Mark Hamburg wrote:
> [...]
> > By what syntactic rule would you recognize that this was the correct
> > parsing? You have no type information available about any of the keywords
> > above. How would you distinguish between a parsing of:
> 
> You don't. Functions are always left-associative.

You mean right-associative.

So, supposing you write:

function foo()
	sparkly(large)
	green(object)
end

Bearing in mind that (apart from one little hitch) Lua doesn't
consider newlines significant, your Lua would interpret that the way
the current Lua interprets:

function foo()
	sparkly(large(green(object)))
end

How would you get the expected behaviour of the first case? Introduce
semicolons? Make newlines end a statement? Now we're talking about
making a completely new language...

OTOH, I see no reason why one couldn't write:

sparkly function() ... end

or:

sparkly true

or:

x = {
	foo()
	bar()
	baz()
}

Especially the last of those (I think sol has this?) would be quite
useful for data description. Maybe these are absent because of the
arbitrary-line-noise-is-valid-perl argument? If so, I don't think I
agree. Strong testing and all... (not that I actually agree with that
article)

Anyway, good syntaxes are surprising hard to design, and there are
always compromises. Requirement of brackets in some cases is one of
Lua's.

-- Jamie Webb