Re: [boost] [range] Should the result of irange satisfy the BidirectionalRangeConcept?

Subject: Re: [boost] [range] Should the result of irange satisfy the BidirectionalRangeConcept?
From: Neil Groves (neil_at_[hidden])
Date: 2011-03-13 18:45:23

On Sat, Mar 12, 2011 at 7:26 PM, Vicente Botet <vicente.botet_at_[hidden]>wrote:

> Hi,
>
> I'm trying to define an enum_range and I based my implementation on the one
> from irange.
>
> While writing my test I started just by checking that my enum_range
> satisfied the RandomAccessRangeConcept requirements and there were some
> errors with BidirectionalRangeConcept. I replaced by the result type of the
> irange function and the errors appear also there.
>
> It seems that there is an issue concerning the category tags
>
> ..\..\../boost/concept_check.hpp:212:13: error: conversion from
>
> 'boost::detail::iterator_category_with_traversal&lt;std::input_iterator_tag,
> boost::random_access_traversal_tag&gt;' to non-scalar type
> 'std::bidirectional_iterator_tag' requested
>
> Am I missing something?
>
>
You are almost certainly not missing anything. I have to look at some other
iterator traversal category related issues particularly with the demotion of
tags. I think there are some interoperability issues in this area. I need to
make some time to review these issues, and put in new test cases. I'm a bit
snowed under at work, but will try and get something done quickly. This
issue is clearly affecting a few people now.

> Best,
> Vicente
>
>
Thanks for the report,
Neil Groves