range and algorithm-related stuff

[Andrei Alexandrescu]

Christopher Wright wrote:
> Andrei Alexandrescu wrote:
>> Sergey Gromov wrote:
>>> Sat, 24 Jan 2009 17:09:07 -0800, Andrei Alexandrescu wrote:
>>>
>>>> I'm working on the new range stuff and the range-based algorithm. In 
>>>> all likelihood, you all might be pleased with the results.
>>>>
>>>> I wanted to gauge opinions on a couple of issues. One is, should the 
>>>> empty() member function for ranges be const? On the face of it it 
>>>> should, but I don't want that to be a hindrance. I presume non-const 
>>>> empty might be necessary sometimes, e.g. figuring out if a stream is 
>>>> empty effectively means fetching an element off it.
>>>
>>> I have a hard time imagining a use for a const range.
>>
>> Read-only arrays for example.
> 
> A range is essentially an iterator. It has to change its internal state 
> to move to the next element. So a const range will not allow you to 
> iterate over the members of a const array. It will allow you to iterate 
> over a single element, either once or an infinite number of times.

A range offering random access can give me any element without the range 
actually changing.

> You could change ranges to have "Range next()" rather than "void 
> next()", and that would allow you to use a const Range for reasons other 
> than checking whether it's empty. Iterating over a range would then look 
> like:
> for (auto range = getRange(); !range.empty; range = range.next) {}
> 
> I don't see much purpose to this. If you want polymorphic ranges, you're 
> going to use a class for ranges. This will incur a lot of allocations, 
> which will be dog slow. The current design would only allocate once if 
> your range is a class.

I agree. Range.next currently mutates the range in-place.


Andrei