From: Gavin Sinclair <gsinclair@...>
Date: 2002-12-09T16:46:33+09:00
Subject: Re: [YANQ] -yet another (ruby) newbie question -string#concat

From: "Pe�a, Botp" <botp@delmonte-phil.com>

> > Not every method that modifies the receiver has a ! -- for
> > example, String#replace.  (If there were a replace!, then the
> > original replace would just be dup :-)  I guess to me
> > 'concat' has the same inherently change-rendering implication
>
> Then #replace method is another thing I would avoid since the syntax/naming
> has implications or assumptions. Syntax/Naming should be unambiguous as much
> as possible, imho... Why can we not standardize on suffixing "!" on methods
> that modifies objects? Are they too many? Though, this is not a big problem
> for me really, as of now, since I'm actually avoiding them.. :-)


Not every destructive method is marked with a ! because ......

 ... I believe the benefit of the "!" notiation is to distinguish between
methods that are similar in semantics but have the subtle difference that one
modifies the recevier and the other does not.

Thus, we are enriched by having:

    String#gsub
    String#gsub!

    Array#uniq
    Array#uniq!

    Kernel::chomp
    Kernel::chomp!

for example.

If we marked *every* destructive method with "!" then we would lose that
quality about "!" which so distinguishes it as a language feature in the first
place.  As David wrote, Array#push! and String#replace! etc. would get a bit
tiresome.

Also note that most library classes do not contain many, if any "!" methods,
because they do not need to make the distinction.  A method either modifies the
receiver or it doesn't.  There's no use having one that does and one that
doesn't.

"?" methods are used in most Ruby code, though, because there is no distinction
that can possibly be made between "Klass#method" and "Klass#method?".  A method
is either a predicate or it's not.  "?" is just a very nice coding style that
I'm very grateful for.

Gavin