Retrieving the traversed range

[Steven Schveighoffer]

On Wed, 25 Aug 2010 09:10:18 -0400, Andrei Alexandrescu  
<SeeWebsiteForEmail at erdani.org> wrote:

> On 8/25/10 1:08 PDT, Peter Alexander wrote:
>> Maybe I'm missing something, but I can't think of anyway *at all* to do
>> this generically.
>>
>> Lets say I have some arbitrary bidirectional range, R, and I want to
>> find the first element that satisfies some predicate. After that, I want
>> to reverse the part of the range up to that element.
>>
>> Essentially, I'd like to do something along the lines of:
>>
>> reverse(until!(pred)(R));
>>
>> but Until is (correctly) not bidirectional, so I can't do that.
>>
>> Use indexing and slicing is not an option because the range isn't
>> necessary random-access.
>>
>> This isn't about what std.algorithm and std.range can do -- I can't even
>> think of a way to do this using primitive range operations.
>>
>> Also note that popping off from the back of the range is not an option
>> either because the range could be arbitrarily long and the predicate is
>> usually satisfied very early in the range.
>>
>> Thanks in advance.
>
> This is an annoying limitation of bidirectional ranges that we don't  
> have a solid solution to. Let me put on the table the unacceptable  
> solution as a baseline:
>
> auto n = walkLength(r) - walkLength(until!pred(r));
> popBackN(r, n);
> reverse(r);
>
> Of all ranges, bidirectional ranges suffer of this limitation because  
> they clearly have two underlying "ends" that are inaccessible in  
> separation from the range.

*cough* cursors *cough* ;)

-Steve