From: Brian Candler <B.Candler@...>
Date: 2002-10-15T18:39:01+09:00
Subject: Re: 0 is true - how'bout "least surprises" ;-P (Was: Re: & operation and if statment)

On Tue, Oct 15, 2002 at 06:43:53AM +0900, Szabolcs Szasz wrote:
> >  > Don't forget that 0 is true.
> > 
> > Duh, thank you, I had compelely forgotten that.
> 
> Er... Not that I couldn't fix my scripts (my server monitor 
> just stopped alerting me of a recent downtime exactly for 
> this mistake...), but, Matz, is "if 0" being the same as 
> "if true" your final word on this? ;-)

One of the things I find annoying about Perl is having to write everywhere
    if (defined $x)
or  if (exists $x{$y})

because 'if ($x)' introduces subtle special-case bugs (where $x is a string
containing exactly "0", for example) - and I see many examples of code which
is broken in this way. IMO those surprises, in 0.01% of cases, are far more
annoying and difficult to track down than something which persistently fails
simply because the language does not do what a 'C' programmer would expect.

Also, bitwise tests don't occur very frequently. I'd be quite happy to write
    puts "yippee" if (data & 0x80) == 0
and it makes the intention very clear.

Perhaps it would be helpful if ruby -w generated a warning if a Fixnum is
used as a truth value; that would force you to write
    if myint != 0
(However, if you don't know the type of an object in advance you could
also be forced to write 'unless param.nil?' instead of 'if param' to
avoid this warning, which takes us back to the Perl situation)

Regards,

Brian.