[CMake] CMake still broken post-2.8.1
Phil Smith
phil at voltage.com
Sat Jan 21 12:24:01 EST 2012
Ah HAH -- you're right, sort of, though I don't understand why it's this way. With the code added that you suggested:
GOOD
!!!! C:/Program Files/Regina/regina.exe cc.rex dcc.exe CMakeCCompilerId.c
-- arg=[COMMAND]
-- arg=[C:/Program Files/Regina/regina.exe]
-- arg=[cc.rex dcc.exe]
-- arg=[CMakeCCompilerId.c]
-- arg=[WORKING_DIRECTORY]
-- arg=[C:/SVN/zFPE/CMakeFiles/CompilerIdC]
-- arg=[OUTPUT_VARIABLE]
-- arg=[CMAKE_C_COMPILER_ID_OUTPUT]
-- arg=[ERROR_VARIABLE]
-- arg=[CMAKE_C_COMPILER_ID_OUTPUT]
-- arg=[RESULT_VARIABLE]
-- arg=[CMAKE_C_COMPILER_ID_RESULT]
BAD
!!!! c:/Program Files/Regina/regina.exe cc.rex;dcc.exe CMakeCCompilerId.c
-- arg=[COMMAND]
-- arg=[c:/Program Files/Regina/regina.exe]
-- arg=[cc.rex]
-- arg=[dcc.exe]
-- arg=[CMakeCCompilerId.c]
-- arg=[WORKING_DIRECTORY]
-- arg=[C:/SVN/zFPE/CMakeFiles/CompilerIdC]
-- arg=[OUTPUT_VARIABLE]
-- arg=[CMAKE_C_COMPILER_ID_OUTPUT]
-- arg=[ERROR_VARIABLE]
-- arg=[CMAKE_C_COMPILER_ID_OUTPUT]
-- arg=[RESULT_VARIABLE]
-- arg=[CMAKE_C_COMPILER_ID_RESULT]
But confusion continues. I changed cc.rex to echo its arguments on entry and its return code on exit to a log, and:
GOOD:
dcc.exe -o CMakeFiles/cmTryCompileExec.dir/testCCompiler.c.o -c C:/SVN/zFPE/CMakeFiles/CMakeTmp/testCCompiler.c
Exiting cc.rex, rc=0
dcc.exe -o CMakeFiles/cmTryCompileExec.dir/CMakeCCompilerABI.c.o -c C:/Program Files (x86)/CMake 2.8/share/cmake-2.8/Modules/CMakeCCompilerABI.c
Exiting cc.rex, rc=1
dcxx.exe -o CMakeFiles/cmTryCompileExec.dir/testCXXCompiler.cxx.o -c C:/SVN/zFPE/CMakeFiles/CMakeTmp/testCXXCompiler.cxx
Exiting cc.rex, rc=0
dcxx.exe -o CMakeFiles/cmTryCompileExec.dir/CMakeCXXCompilerABI.cpp.o -c C:/Program Files (x86)/CMake 2.8/share/cmake-2.8/Modules/CMakeCXXCompilerABI.cpp
Exiting cc.rex, rc=1
dcc.exe -o CMakeFiles/cmTryCompileExec.dir/CheckIncludeFile.c.o -c C:/SVN/zFPE/CMakeFiles/CMakeTmp/CheckIncludeFile.c
Exiting cc.rex, rc=0
dcc.exe -o CMakeFiles/cmTryCompileExec.dir/CheckIncludeFile.c.o -c C:/SVN/zFPE/CMakeFiles/CMakeTmp/CheckIncludeFile.c
Exiting cc.rex, rc=0
dcc.exe -o CMakeFiles/cmTryCompileExec.dir/CheckIncludeFile.c.o -c C:/SVN/zFPE/CMakeFiles/CMakeTmp/CheckIncludeFile.c
Exiting cc.rex, rc=0
dcc.exe -o CMakeFiles/cmTryCompileExec.dir/SIZEOF_UNSIGNED_LONG_LONG.c.o -c C:/SVN/zFPE/CMakeFiles/CheckTypeSize/SIZEOF_UNSIGNED_LONG_LONG.c
Exiting cc.rex, rc=0
dcc.exe -o CMakeFiles/cmTryCompileExec.dir/SIZEOF_UNSIGNED_LONG.c.o -c C:/SVN/zFPE/CMakeFiles/CheckTypeSize/SIZEOF_UNSIGNED_LONG.c
Exiting cc.rex, rc=0
dcc.exe -o CMakeFiles/cmTryCompileExec.dir/SIZEOF_UNSIGNED_INT.c.o -c C:/SVN/zFPE/CMakeFiles/CheckTypeSize/SIZEOF_UNSIGNED_INT.c
Exiting cc.rex, rc=0
BAD:
dcc.exe CMakeCCompilerId.c
So arguably this is a bug in my cc.rex, in that it's not noticing the missing arguments. But of course, "it worked before". Also note that cc.rex *never gets invoked against CMakeCCompilerId* in the GOOD case.
The more I look at this, the less I understand it...
-----Original Message-----
From: Phil Smith
Sent: Friday, January 20, 2012 4:37 PM
To: 'Brad King'
Cc: Bill Hoffman; cmake at cmake.org
Subject: RE: [CMake] CMake still broken post-2.8.1
I'm confused because neither of your examples has the semicolon, but what we need is:
"regina.exe" "cc.rex" "dcc.exe" "CMakeCCompilerId.c"
But it sounds like you're saying that we're not actually getting invoked as:
cc.rex;dcc.exe CMakeCCompilerId.c
The problem is, I can't tell what we ARE getting invoked as. The symptoms match regina being invoked with no operands, but I can't seem to prove that.
If I run with --trace and --debug-output, the last of the output is:
C:/Program Files (x86)/CMake 2.8/share/cmake-2.8/Modules/CMakeDetermineCompilerId.cmake(97): EXECUTE_PROCESS(COMMAND ${CMAKE_${lang}_COMPILER} ${CMAKE_${lang}_
COMPILER_ID_ARG1} ${CMAKE_${lang}_COMPILER_ID_FLAGS_LIST} ${testflags} ${src} WORKING_DIRECTORY ${CMAKE_${lang}_COMPILER_ID_DIR} OUTPUT_VARIABLE CMAKE_${lang}_C
OMPILER_ID_OUTPUT ERROR_VARIABLE CMAKE_${lang}_COMPILER_ID_OUTPUT RESULT_VARIABLE CMAKE_${lang}_COMPILER_ID_RESULT )
Called from: [2] C:/Program Files (x86)/CMake 2.8/share/cmake-2.8/Modules/CMakeDetermineCCompiler.cmake
[1] C:/SVN/zFPE/CMakeLists.txt
Does that tell us anything?
-----Original Message-----
From: Brad King [mailto:brad.king at kitware.com]
Sent: Friday, January 20, 2012 3:50 PM
To: Phil Smith
Cc: Bill Hoffman; cmake at cmake.org
Subject: Re: [CMake] CMake still broken post-2.8.1
On October 24, 2011 6:08 PM Bill Hoffman wrote:
> So, right at the end it is doing this:
>
> C:/Program Files (x86)/CMake 2.8/share/cmake-2.8/Modules/CMakeDetermineCompilerId.cmake(96):
> EXECUTE_PROCESS(COMMAND ${CMAKE_${lang}_COMPILER} ${CMAKE_${lang}_COMPILER_ID_ARG1}
> ${CMAKE_${lang}_COMPILER_ID_FLAGS_LIST} ${testflags} ${src}
> WORKING_DIRECTORY ${CMAKE_${lang}_COMPILER_ID_DIR}
> OUTPUT_VARIABLE CMAKE_${lang}_COMPILER_ID_OUTPUT
> ERROR_VARIABLE CMAKE_${lang}_COMPILER_ID_OUTPUT
> RESULT_VARIABLE CMAKE_${lang}_COMPILER_ID_RESULT )
[snip]
> message("${CMAKE_${lang}_COMPILER} ${CMAKE_${lang}_COMPILER_ID_ARG1} ${CMAKE_${lang}_COMPILER_ID_FLAGS_LIST} ${testflags} ${src}"
[snip]
On 10/24/2011 6:38 PM, Phil Smith wrote:
> Ah, ok. With 2.8.1:
>
> C:/Program Files/Regina/regina.exe cc.rex dcc.exe CMakeCCompilerId.c
>
> With 2.8.6:
> C:/Program Files/Regina/regina.exe cc.rex;dcc.exe CMakeCCompilerId.c
>
> Note the semicolon; where'd it come from? It's not in the toolchain file:
I did not notice previously that the reported ; is coming from a message()
command in a string and not a real command line that is invoked. The
2.8.1 output line misrepresents the actual command that is running.
What you mean by:
> SET(CMAKE_C_COMPILER "regina.exe" "cc.rex dcc.exe")
> SET(CMAKE_CXX_COMPILER "regina.exe" "cc.rex dcxx.exe")
is that you want the compiler to be invoked as
"regina.exe" "cc.rex dcc.exe" "CMakeCCompilerId.c"
where the second piece puts both files in a *single* argument, right?
Much of the discussion in this thread made it appear that you wanted
"regina.exe" "cc.rex" "dcc.exe" "CMakeCCompilerId.c"
but that is what the above-mentioned execute_process command ends up
running in 2.8.6 and causing the hang.
Correct?
-Brad
More information about the CMake
mailing list