The temporary variable would not have to store the execution state of the coroutine, only the yielded value. Offering a broken postfix increment is very dangerous and should be avoided. Also, if I can’t count on InputIterator features being supported, the following is a lie:

 

class iterator : public std::iterator< std::input_iterator_tag, typename remove_reference< R >::type >

 

(taken from asymmetric_coroutine.hpp)

 

--

Johannes S. Mueller-Roemer, MSc

Wiss. Mitarbeiter - Interactive Engineering Technologies (IET)

 

Fraunhofer-Institut für Graphische Datenverarbeitung IGD

Fraunhoferstr. 5  |  64283 Darmstadt  |  Germany

Tel +49 6151 155-606  |  Fax +49 6151 155-139

johannes.mueller-roemer@igd.fraunhofer.de  |  www.igd.fraunhofer.de

 

From: Boost-users [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Oliver Kowalke
Sent: Monday, February 16, 2015 15:28
To: boost-users@lists.boost.org
Subject: Re: [Boost-users] [context] boost::coroutines::asymmetric_coroutine<T>::pull_type iterator does not properly model InputIterator

 

coroutines behave different than ordinary containers - even a temp var, created inside iterator::operator++( int), will influence the state of the attached coroutine (increment operator)

you can't count on the feature set described for input iterators in the standard

 

2015-02-16 15:00 GMT+01:00 Mueller-Roemer, Johannes Sebastian <Johannes.Sebastian.Mueller-Roemer@igd.fraunhofer.de>:

If that is „as expected“ it still violates the requirements for an InputIterator. See the last row in Table 107 in 24.2.3 in the standard http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2014/n4296.pdf

As-is the current “iterator” is not a valid iterator of any kind as InputIterator is already the weakest requirement.

 

Example of how it should work (using an istream_iterator, which is an InputIterator):

#include <iostream>

#include <sstream>

#include <iterator>

 

auto main(int, char **) -> int

{

      std::istringstream str("Hello , World !");

      for(auto it = std::istream_iterator<std::string>(str); it != std::istream_iterator<std::string>();)

            std::cout << *it++;

      std::cout << "\n";

}

 

Which outputs (as expected):

Hello,World!

 

The issue can be solved by storing a value in the iterator or returning a proxy object which dereferences to the original value.

--

Johannes S. Mueller-Roemer, MSc

Wiss. Mitarbeiter - Interactive Engineering Technologies (IET)

 

Fraunhofer-Institut für Graphische Datenverarbeitung IGD

Fraunhoferstr. 5  |  64283 Darmstadt  |  Germany

Tel +49 6151 155-606  |  Fax +49 6151 155-139

johannes.mueller-roemer@igd.fraunhofer.de  |  www.igd.fraunhofer.de

 

From: Boost-users [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Oliver Kowalke
Sent: Monday, February 16, 2015 14:20
To: boost-users@lists.boost.org
Subject: Re: [Boost-users] [context] boost::coroutines::asymmetric_coroutine<T>::pull_type iterator does not properly model InputIterator

 

2015-02-16 13:46 GMT+01:00 Mueller-Roemer, Johannes Sebastian <Johannes.Sebastian.Mueller-Roemer@igd.fraunhofer.de>:

Compiling is not the issue.

 

it works as expected

 

Furthermore your example does not test the issue I mentioned, as it does not use *it++;

 

operator++ has higher precedence than operator* -> each increment triggers an jump/switch operation -> last increment invalidates coroutine

 


_______________________________________________
Boost-users mailing list
Boost-users@lists.boost.org
http://lists.boost.org/mailman/listinfo.cgi/boost-users