From: Brian Candler <B.Candler@...>
Date: 2005-07-19T18:14:38+09:00
Subject: Re: What does this construct mean?

On Tue, Jul 19, 2005 at 02:50:54AM +0900, Casper wrote:
> I guess two reasons that I had difficult figuring out are (1) Ruby
> allows executable stuff to go where I expect to see declarations, but
> moreover (2) Ruby allows method calls to have their arguments
> delineated with or without parentheses. In particular, I don't
> understand the why you would want this. Is there a good motivation to
> allow it?

Probably because of the first program people see:

    puts "Hello, world!"

versus:

    puts("Hello, world!")

However, even this simple example highlights a problem. Insert a space
between puts and ( and you get a warning:

    $ ruby -we 'puts ("Hello")' 
    -e:1: warning: (...) interpreted as grouped expression
    Hello

    $ ruby -we 'puts ("Hello","world")'
    -e:1: warning: don't put space before argument parentheses
    Hello
    world

Even though (a,b,c) is not a valid expression when standing alone, the Ruby
interpreter has difficulty deciding whether (a) is an expression wrapped in
parentheses, or a method call argument list wrapped in parentheses.

To see where it makes a difference, try these:

    $ ruby -e 'puts ("a"), ("b")'              
    $ ruby -e 'puts("a"), ("b")'

    $ ruby -e 'puts (3+4).abs'
    $ ruby -e 'puts(3+4).abs'

I think the space-sensitive behaviour is driven by this last case: i.e. that
"method1 (expr).method2" should be parsed as "method1( (expr).method2 )"
whereas "method1(expr).method2" is "(method1(expr)).method2"

Regards,

Brian.