From: Brian Candler <B.Candler@...>
Date: 2004-11-06T22:38:16+09:00
Subject: Re: argument parenths

> |Any chance the parenths/space requirement will go away in 2.0?
> |
> |  method  (1, 2, 3)
> 
> Not sure.  Only if I can make yacc rules smart enough to parse/detect
> 
>   method (arg1), arg2

Are you talking about the warning you get if you do
  method (1, 2, 3)
instead of
  method(1, 2, 3)
?

I wonder if this is related to the bracket ambiguity, i.e.

  puts (a+b).abs        =>        puts((a+b).abs)
  puts(a+b).abs         =>        (puts(a+b)).abs

Neither of these generates a warning (1.8.2p2), but they do different things
depending on whether the space is there or not.

Trying a few combinations with more than one argument:

1.  puts (a+b), c    works - no warning is generated
2.  puts(a+b), c     fails
3.  puts(a+b, c)     works - no warning
4.  puts (a+b, c)    works - but generates a warning
5.  puts (a+b)       works - no warning
6.  puts(a+b)        works - no warning

So I'm not exactly sure why case 4 generates a warning, but case 5 doesn't.
In case 4, the brackets contain a comma, so surely it must be an argument
list, and therefore must be associated with the method name before it?

The example "method (arg1), arg2" seems to be an example of case 1, and that
works just fine under 1.8.2-p2

There is an ambiguity for

   foo(bar (a+b), c)

which generates
warning: parenthesize argument(s) for future version

but it seems to resolve it as

  foo(bar((a+b),c))

which I suppose is as reasonable a guess as any.

Regards,

Brian.