Hi --

On Mon, 20 Jul 2009, NARUSE, Yui wrote:

> David A. Black wrote:
>> On Mon, 20 Jul 2009, NARUSE, Yui wrote:
>>> When the number of argument for Hash[] is 1,
>>> the argument must be a hash (or hash like object) or an array (or
>>> array like).
>>>
>>> And the array object must be an array of two or one length array.
>>> Your code is an array of integer, so the code returns the empty hash.
>>
>> I'm not sure I understand the reasoning behind that. It seems awfully
>> complicated and hard to follow:
>>
>> snip<
>>
>> I'm not clear on the underlying design principle, I guess. The 1.8
>> version seems like it was easier to read; an odd number of arguments
>> was wrong, which makes sense (to me) for a hash.
>
> The motivation of thie spec is for an array of key-value arrays.
>
>>> Hash[[[:a,1], [:b,2], [:c]]]
> => {:a=>1, :b=>2, :c=>nil}
>
> So arrays beyond this will be empty.
>
>> Wouldn't it be clearer if Hash[[1,2]] was an error?
>
> How about following,
>
>>> Hash[[[:a,1], [:b,2], [:c], nil]]
> => {:a=>1, :b=>2, :c=>nil}
>
> But raise if the content of argument array is not neither array nor nil,
> it maybe adoptable without compatibility problem.

It still seems like Hash[[1,2]] is just a hash with an odd number of
arguments. I can see why some array interpretation might be useful,
with nested arrays. But if someone writes Hash[one_level_array],
it's more like that it's a mistake, than that the person is trying to
construct an empty hash. So I think it would be more useful to have it
raise an exception.


David

-- 
David A. Black / Ruby Power and Light, LLC
Ruby/Rails consulting & training: http://www.rubypal.com
Now available: The Well-Grounded Rubyist (http://manning.com/black2)
Training! Intro to Ruby, with Black & Kastner, September 14-17
(More info: http://rubyurl.com/vmzN)