Errors compiling Boost.Python library on Solaris
Hi, I'm new to this list and to Boost. I'm interested in using Boost to wrap C++ objects in Python. The environment is Solaris 10, x86. I did the following: Downloaded boost_1_33_1.tar and extracted it to /opt/boost.
setenv PYTHON_ROOT /usr/local cd to /opt/boost/boost_1_33_1/libs/python/build bjam
I get lots of error messages, repeated throughout the compile and tend to be grouped and refer to the same line: '->' cannot appear in a constant-expression '&' cannot appear in a constant-expression template argument 1 is invalid value is not a member of '<declaration errors>' ...failed gcc-C++-action ... is also associated with the above If I "bjam" from /opt/boost/boost_1_33_1, I get the same errors when it gets to the python library. I would most appreciate any help with this! Regards, Chris Botos
"chris botos"
Hi,
I'm new to this list and to Boost. I'm interested in using Boost to wrap C++ objects in Python. The environment is Solaris 10, x86. I did the following:
Downloaded boost_1_33_1.tar and extracted it to /opt/boost.
setenv PYTHON_ROOT /usr/local cd to /opt/boost/boost_1_33_1/libs/python/build bjam
I get lots of error messages, repeated throughout the compile and tend to be grouped and refer to the same line: '->' cannot appear in a constant-expression '&' cannot appear in a constant-expression template argument 1 is invalid value is not a member of '<declaration errors>' ...failed gcc-C++-action ... is also associated with the above
If I "bjam" from /opt/boost/boost_1_33_1, I get the same errors when it gets to the python library.
I would most appreciate any help with this!
http://engineering.meta-comm.com/boost-regression/CVS-RC_1_33_0/user/python.... shows that the library worked on SunOS with gcc-3.4.3 Details about that tester's setup are at: http://engineering.meta-comm.com/boost-regression/CVS-RC_1_33_0/CalebEpstein... HTH, -- Dave Abrahams Boost Consulting www.boost-consulting.com
On 10/18/06, chris botos
I'm new to this list and to Boost. I'm interested in using Boost to wrap C++ objects in Python. The environment is Solaris 10, x86. I did the following:
'->' cannot appear in a constant-expression '&' cannot appear in a constant-expression template argument 1 is invalid value is not a member of '<declaration errors>' ...failed gcc-C++-action ... is also associated with the above
What version of Solaris and gcc are you using? Can you include *actual* error messages with actual filenames and line numbers to help diagnose the problem? I'm currently running the RC-1_34_0 and CVS-HEAD regression tests on a Solaris/x86 system with gcc 4.1.1. I am pretty sure gcc 3.4.x will work fine as well, but 3.3.x has issues with wide strings and I/O streams unless you are on Solaris 10 IIRC. Here's my current config: http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/su... The links Dave gives for Boost 1.33.0 are from a Solaris/SPARC machine to which I no longer have access. -- Caleb Epstein
On 11/2/06, Caleb Epstein
On 10/18/06, chris botos
wrote: I'm new to this list and to Boost. I'm interested in using Boost to wrap C++ objects in Python. The environment is Solaris 10, x86. I did the following:
'->' cannot appear in a constant-expression '&' cannot appear in a constant-expression template argument 1 is invalid value is not a member of '<declaration errors>' ...failed gcc-C++-action ... is also associated with the above
Caleb and Dave, Thanks so much for your replies! I didn't see them until this weekend, so I hadn't responded sooner. Being unable to get Boost to compile, and needing to get some Python extensions working right away, I started using SWIG, which so far has met my basic needs. What version of Solaris and gcc are you using? Solaris 10/x86 gcc 3.4.3 Can you include
*actual* error messages with actual filenames and line numbers to help diagnose the problem?
I attached the full output from bjam. I'm not sure when I will be able to get back to Boost. It may depend on whether I run into SWIG limitations. Regards, Chris
On 11/13/06, chris botos
On 11/14/06, Caleb Epstein
Not sure if it helps, but the compiler seems to be complaining about this line:
BOOST_STATIC_ASSERT(BOOST_PYTHON_OFFSETOF(rvalue_from_python_storage<T>,stage1) == 0);
Where Solaris appears to define offsetof() in /usr/include/iso/stddef_iso.h as:
#if __cplusplus >= 199711L #define offsetof(s, m) (std::size_t)(&(((s *)0)->m)) #else #define offsetof(s, m) (size_t)(&(((s *)0)->m)) #endif
Not sure if there is much that I can do to help aside from saying "it works for me" using gcc 4.1.1.
Thanks for looking at this and the info, Caleb. I'll see about about upgrading gcc and giving it another try. - Chris
participants (3)
-
Caleb Epstein
-
chris botos
-
David Abrahams