Arrays are sufficient for ArrayLists? Really??

[Steven Schveighoffer]

On Mon, 16 May 2011 16:14:33 -0400, Timon Gehr <timon.gehr at gmx.ch> wrote:

> Steven Schveighoffer wrote:
>> On Mon, 16 May 2011 15:52:23 -0400, Mehrdad <wfunction at hotmail.com>  
>> wrote:
>>
>>> On 5/16/2011 12:53 PM, Timon Gehr wrote:
>>>>> In fact I even need to take that back. In order to work correctly,  
>>>>> the
>>>>> function would have to iterate downwards. It _is_ indeed buggy, and I
>>>>> should stop emitting opinions when I'm short on time...
>>>>>
>>>>> Andrei
>>>>
>>>> Whoops, you are right:
>>>>
>>>> void removeAt(T)(ref T[] arr, size_t index)
>>>> {
>>>>     foreach (i, ref item; retro(arr[1 .. index+1]))
>>>>          item = arr[i - 1];
>>>>      arr = arr[1 .. $];
>>>> }
>>>>
>>>> Timon
>>>
>>> Wouldn't that stomp on the super-slice of arr, though?
>>>
>>> As was pointed out on SO, the problem is actually always there: if
>>> someone passes arr[0 .. $] to a function, it will look as if the
>>> original array was passed, although it wasn't. Seems like it's a lot
>>> uglier than I'd thought...
>>
>> arr[0..$] shouldn't be an lvalue (it is, but I think that's a bug).
>> Therefore, you shouldn't be able to pass it as a ref argument (is this  
>> in
>> bugzilla?).
>>
>> -Steve
>
> I think it is a little bit more subtle than "not lvalue", because you  
> actually
> have to be able to do something like:
>
> arr[0..2] = arr[3..5];
>
> Where a slice appears on the left hand side of the assignment. (also  
> a[]=b[]).

ref is not required in those cases, the array can be passed by value (and  
by value, I mean the pointer and length, not the data).

You only need the array to be an lvalue when you want the passed-in  
array's pointer and length to be modified.

-Steve