A few issues in the policy-refactor about websocketpp

[zaphoyd]

Regarding the friend point. Do you know what the correct way to do this pre-c++11 is? The commented out definitions are flagged by LLVM/Clang with a C++11 only warning. The uncommented lines work fine in GCC 4.2+ and LLVM/Clang without C++11 enabled. Which version of Visual Studio are you using and do you know what it's C++11 status as far as extended friend declarations is?

[megabyte1024]

Do you know what the correct way to do this pre-c++11 is?
Unfortunately no. I have not ever worked with templates on such deep level. It's a good chance. I tried to find a universal solution by reviewing boost sources, but I have no time to "dig" deeper. My proposal to leave this place as it is now, till the better times.

I use VS2010 and I am not aware about the status of extended friend declarations. I am going to search in this direction when I have time.

[megabyte1024]

A couple of new topics to fix.

1. In my previous fix #49 was a mistake. The diff for the data.cpp file contained the original line in old and new case. Here is a fix

diff --git a/src/messages/data.cpp b/src/messages/data.cpp

•        std::max(new_size, static_cast<uint64_t>(2*m_payload.capacity()))
  

•        std::max<uint64_t>(new_size, static_cast<uint64_t>(2*m_payload.capacity()))
  

1. The Windows header files already have the ERROR define, and as result there is a compilation error one lines something like that

m_endpoint.elog().at(log::elevel::ERROR)

To fix it is necessary to rename the log::elevel::ERROR constant to something different. My patch is

diff --git a/src/logger/logger.hpp b/src/logger/logger.hpp

• static const value ERROR = 0x10;
• static const value _ERROR = 0x10;

plus renaming of all ERROR to _ERROR.