From: Robert Dober <robert.dober@...>
Date: 2008-07-22T04:47:55+09:00
Subject: [ruby-core:17904] Re: duping the NilClass

On Mon, Jul 21, 2008 at 6:58 PM, David A. Black <dblack@rubypal.com> wrote:
> On Tue, 22 Jul 2008, Robert Dober wrote:
>
>> On Mon, Jul 21, 2008 at 5:18 PM, David A. Black <dblack@rubypal.com>
>> wrote:
>> <snip>
>>>
>>> I'm afraid I don't follow that at all. What does it have to do with
>>> other ! methods?
>>
>> Well I interpreted your statement as follows:
>> dup is a bad name for Symbol or NilClass as it does not really create
>> a new symbol or nil
>> thus I call it dup!
>> By extension that would mean that e.g.
>> gsub! is called gsub! because it does not really gsub
>>
>> Did I read you incorrectly so far?
>
> Yes. There's no general "___ does not really ___" rule. The only
> general rule about ! methods is the idea of "danger", which can take
> many forms.
>
>> There is already some disagreement about the ! suffix so far, your new
>> interpretation of ! meaning dangerous
>
> It's absolutely not new; Matz has always characterized it that way
> ("dangerous").
>
>> in the sense that it does something slightly different of what might
>> be expected might add to the confusion of
>> what ! should mean.
>> IMHO it should mean a modifying version of a non modifying version of
>> a method. But there is not much agreement on this :(.
>
> There's been a lot of misunderstanding and second-guessing of Matz,
> but he's always been consistent in his description.
>
>> However there seems agreement that xxx! is a more "dangerous" version
>> of xxx and in our case there would
>> be dup! but no dup for e.g. NilClass.
>>
>> Are there other classes, modules, singletons with an xxx! without an
>> xxx <space> ?
>
I do not want to bother this list with my limited ability of
expressing myself in English. Obviously I am quite incompetent to say
what I want to say, unless you are playing tricks on me David, but I
guess you would not do this. Believe me, your replies do not at all
fit what I wanted to say, just let us forget (can we discuss this in
French, well actually they do not understand me much lately either)
... :(
> No; ! methods only make sense in pairs with non-! methods. See my blog
> post (cited earlier). Unfortunately, there's a trend toward sticking !
> on methods that aren't in pairs. It makes it meaningless.
This is exactly what I said though, your idea would break this rule as
there would be NilClass#dup! but no NilClass#dup.

Maybe this of my many observations in this thread can be considered
useful at least?


> Must go -- my students are back from lunch :-)
Hey you are lucky, here in France, an instructor is supposed to share
lunch with her class!!!

Cheers
Robert