No errors, no CMakeError.log -- just null values for the SIZEOF_LONG_LONG et al.
It sets CMAKE_C_COMPILER_ARG1 in CMakeDetermineCCompiler.cmake -- but inside:
IF(NOT CMAKE_C_COMPILER)
...so it doesn't do it if you explicitly specify a C compiler, a la:
SET(CMAKE_C_COMPILER "c:/progra~1/dignus/cc.bat -fasciiout")
At least, that's how I read it. After some fumbling, I put a MESSAGE before
and inside that IF block, and only the one before gets triggered.
Related note:
It sure would be nice if --debug-trycompile copied the compiler output to
CMakeError.log. I've been forced to put a
FILE(APPEND ${CMAKE_BINARY_DIR}${CMAKE_FILES_DIRECTORY}/CMakeError.log
"Compiler output:\n${OUTPUT}\n")
in CheckTypeSize.cmake so I can figure out what's failing (not with this, since
it's just not passing the flag, but with other failures before).
...phsiii
-----Original Message-----
From: Alexander Neundorf [mailto:[EMAIL PROTECTED]
Sent: Tuesday, August 05, 2008 5:30 PM
To: Phil Smith
Cc: cmake@cmake.org
Subject: Re: [CMake] FW: cmake newbie questions
On Tuesday 05 August 2008, Phil Smith wrote:
> Nope, that didn't do it. I tried it with and without double quotes around
> the value: no -fascii was passed in either case.
>From a clean build tree, right ?
It should work.
If it doesn't, search for CMAKE_C_COMPILER_ARG1 and check using
MESSAGE(STATUS ...) what happens.
What errors do you get ?
Printed and from CMakeFiles/CMakeError.log ?
> However, I'm tentatively
> declaring victory (very tentatively -- I've had it working before, but for
> the wrong reasons, such as NOT doing a linkedit and having it happen to
> find the "magic string" in the object -- but that's not safe, it could be
> broken across 80-byte records).
>
> I notice that it's creating what looks like it might be a duplicate set of
> directories:
>
> CMakeFiles\CMakeTmp\CMakeFiles\cmTryCompileExec.dir
>
> Does it really mean to do that, or is it intending to have just
> CMakeFiles\cmTryCompileExec.dir
> ? It works, so perhaps machs nicht,
???
"Es macht nichts" ?
The directories are ok as you see them.
Alex
_______________________________________________
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake
