Hello, Boost Phoenix folks –

I sent the message below a few weeks ago, looking for assistance in getting Boost 1.57 to compile with MSVC 2010 and my complex C++ code. I have yet to receive a response. Perhaps it was misdirected; maybe I need to send this to [spirit] or [qi] or [utility] or  …?

In the weeks since I sent this, I have not exactly been idle on the subject. I have installed Boost 1.54, 1.55, and 1.56 on my computer. I have in turn tried each version by just switching out the include and linker locations. So far, I have found out that 1.53, 1.54, and 1.55 compile successfully. However, switching to 1.56 or 1.57, there is no success. The diagnostics for both versions are similar, regarding too few template arguments for a call to [utility] result_of.hpp that  <boost/phoenix/core/detail/preprocessed/function_eval_10.hpp>  is making.

Does anyone have any ideas or suggestions on how to make Boost 1.57 work with MSVC 2010 and my code? I would be grateful.

Thank you,

/Andrew Nordquist

From: Andrew Nordquist
Sent: Thursday, March 19, 2015 12:35 PM
To: 'boost-users@lists.boost.org'
Subject: [Phoenix] Version 1.57 of Phoenix breaks when used with MSVC 2010 compiler?

Hello, folks –

I must be doing something wrong, but I can’t figure it out. I/we have been using Boost in our 20,000+ lines of C++ code for a few years now.

We started with 1.43, went smoothly to 1.53, and although it was bumpy going from MSVC 2008 to MSVC 2010, we are there now.

We are now using MSVC 2010 and Boost 1.53. We use many routines, including Phoenix, Spirit, bind, tie, etc.

I am attempting to migrate to Boost 1.57. Got clobbered with about 400 lines of unexpected diagnostic in the compiler. That is the attached file. In the file, the lines showing the files named “cutoffitem.cpp” and “densitydelegate.cpp” are ours, but what is between these two lines is coming from Boost.  It appears that it could be a C++11 type of problem, but I’m not sure . I read about that potential problem with the [utility] [result_of] class template in the report designated n3276 and in the 1.57 [utility] documentation. I think the major concern with this issue was going to Boost 1.52. In our code, I tried a quick-fix precompiler #define, but it did not work.

Got any ideas? I’m about to go back to version 1.53, at least until I upgrade to Visual Studio 2013.

Thank you,

/Andrew Nordquist

Research Analyst

Southwest Research Institute

San Antonio, TX, USA