From: ptkwt@...1.aracnet.com (Phil Tomson)
Date: 2002-02-20T06:57:55+09:00
Subject: Re: Operator overloading and multiple arguments

In article <20020219145252.C4142@atdesk.com>,
Paul Brannan  <paul@atdesk.com> wrote:
>On Wed, Feb 20, 2002 at 04:17:41AM +0900, Phil Tomson wrote:
>> Anyway, I think it won't be a problem using '<=' for assignment in this 
>> case.  Here's why:
>
>Okay.  But be careful, because many operator have lower precedence than
><=.  If you try to do anything slightly out of the ordinary with your
>assignments, then you will get unexpected results.

Yep.  Already ran into that.  I was going to override 'and' and 'or' so 
that they could be used in equations:

a <= {b or c}
b <= {d and f}

....but 'and' and 'or' have a lower precedence than '<=', so I had to 
change to using '*' and '+', like:

a <= {b + c}
b <= {d * f}

It would kind of be nice to be able to specify a precedence for an 
overloaded operator, but that's a lot to ask for, I suppose.

>
>Some people have suggested using []= with no arguments as a redefinable
>assignment operator.  The pickaxe book lists []= as having the highest
>precedence, but this doesn't make sense, since this would mean 
>
>  a[4] = 1 + 1
>
>would not work as expected.  Is this a bug in the book?

Ruby in a Nutshell lists '[]' as being second only to '::' in precedence.

Phil