[CMake] Using -DOPTION="something=moof" mangles the OPTION value to "something:UNINITIALIZED=moof"

2011-08-29 Thread Marcus Fritzsch
I discovered this behavior on my ubuntu 11.04 amd64 workstation with
cmake 2.8.3 when trying to set an -march= option for gcc:

cmake -DCMAKE_C_FLAGS="-Os -march=native -flto
-freorder-blocks-and-partition" 

was mangled to be

CMAKE_C_FLAGS=-Os -march:UNINITIALIZED=native -flto
-freorder-blocks-and-partition

in the CMakeCache.txt.

I haven't had much luck with the search engine of my choice finding
anything on the issue.
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Using -DOPTION="something=moof" mangles the OPTION value to "something:UNINITIALIZED=moof"

2011-08-29 Thread Marcus Fritzsch
Turns out, using the "verbose" form of the option definition
-DOPTION:TYPE=... does not exhibit the issue.

On 8/29/11, Marcus Fritzsch  wrote:
> I discovered this behavior on my ubuntu 11.04 amd64 workstation with
> cmake 2.8.3 when trying to set an -march= option for gcc:
>
> cmake -DCMAKE_C_FLAGS="-Os -march=native -flto
> -freorder-blocks-and-partition" 
>
> was mangled to be
>
> CMAKE_C_FLAGS=-Os -march:UNINITIALIZED=native -flto
> -freorder-blocks-and-partition
>
> in the CMakeCache.txt.
>
> I haven't had much luck with the search engine of my choice finding
> anything on the issue.
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


[CMake] OpenCL Module?

2011-08-29 Thread Michael Jackson
Does anyone have an "FindOpenCL.cmake" file that they would like to share?
___
Mike Jackson  www.bluequartz.net
Principal Software Engineer   mike.jack...@bluequartz.net 
BlueQuartz Software   Dayton, Ohio

___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] OpenCL Module?

2011-08-29 Thread Michael Wild
On Mon 29 Aug 2011 03:44:55 PM CEST, Michael Jackson wrote:
> Does anyone have an "FindOpenCL.cmake" file that they would like to share?
> ___
> Mike Jackson  www.bluequartz.net
> Principal Software Engineer   mike.jack...@bluequartz.net 
> BlueQuartz Software   Dayton, Ohio
> 
> ___
> Powered by www.kitware.com
> 
> Visit other Kitware open-source projects at 
> http://www.kitware.com/opensource/opensource.html
> 
> Please keep messages on-topic and check the CMake FAQ at: 
> http://www.cmake.org/Wiki/CMake_FAQ
> 
> Follow this link to subscribe/unsubscribe:
> http://www.cmake.org/mailman/listinfo/cmake

Very simplistic, haven't tried it with all possible SDK's on all 
possible platforms (namely I only tried it on Ubuntu with 
nvidia-current-dev). In particular I'm not very happy with the 
PATH_SUFFIXES in the find_library, as this should probably be tailored 
to CMAKE_SIZEOF_VOID_P.

# - Find the OpenCL headers and library
#
# Defines the following if found:
#  OPENCL_FOUND: TRUE if found, FALSE otherwise
#  OPENCL_INCLUDE_DIRS : Include directories for OpenCL
#  OPENCL_LIBRARIES: The libraries to link against
#
# The user can set the OPENCLROOT environment variable to help finding 
OpenCL
# if it is installed in a non-standard place.

set(CMAKE_FIND_FRAMEWORK FIRST)

find_path(OPENCL_INCLUDE_DIR NAMES OpenCL/cl.h CL/cl.h PATHS
   ENV OPENCLROOT
   PATH_SUFFIXES include include/nvidia-current)

find_library(OPENCL_LIBRARY OpenCL PATHS
   ENV OPENCLROOT
   PATH_SUFFIXES lib/x86_64 lib/x64 lib/x86)

mark_as_advanced(OPENCL_INCLUDE_DIR OPENCL_LIBRARY)

include(FindPackageHandleStandardArgs)
find_package_handle_standard_args(OpenCL OPENCL_INCLUDE_DIR 
OPENCL_LIBRARY)

set(OPENCL_INCLUDE_DIRS "${OPENCL_INCLUDE_DIR}")
set(OPENCL_LIBRARIES "${OPENCL_LIBRARY}")
#- EOF

HTH

Michael
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] FindQt4 fails to find QtUiTools in windows cross-environment

2011-08-29 Thread Clinton Stimpson
On Friday, August 26, 2011 10:54:44 am Dominik Schmidt wrote:
> Hey!
> 
> I'm facing an issue with FindQt4 module when cross-compiling from Linux
> to Windows. The QtUiTools are not found on the first run of cmake,
> executing cmake a second time finds it (I attached the diff of
> CMakeCache.txt before and after the second cmake run).
> 
> I suspect the issue is connected to this QTBUG:
> https://bugreports.qt.nokia.com/browse/QTBUG-20498
> 
> The description is only half-true, it's built here, but as a static
> library in contrast to the rest of Qt which is built as dynamic ones.
> 

A fix for this has been put in.
http://www.cmake.org/gitweb?p=cmake.git;a=commit;h=a67be31

-- 
Clinton Stimpson
Elemental Technologies, Inc
Computational Simulation Software, LLC
www.csimsoft.com
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Using -DOPTION="something=moof" mangles the OPTION value to "something:UNINITIALIZED=moof"

2011-08-29 Thread Michael Hertling
On 08/29/2011 02:37 PM, Marcus Fritzsch wrote:
> Turns out, using the "verbose" form of the option definition
> -DOPTION:TYPE=... does not exhibit the issue.
> 
> On 8/29/11, Marcus Fritzsch  wrote:
>> I discovered this behavior on my ubuntu 11.04 amd64 workstation with
>> cmake 2.8.3 when trying to set an -march= option for gcc:
>>
>> cmake -DCMAKE_C_FLAGS="-Os -march=native -flto
>> -freorder-blocks-and-partition" 
>>
>> was mangled to be
>>
>> CMAKE_C_FLAGS=-Os -march:UNINITIALIZED=native -flto
>> -freorder-blocks-and-partition
>>
>> in the CMakeCache.txt.
>>
>> I haven't had much luck with the search engine of my choice finding
>> anything on the issue.

With the following simple CMakeLists.txt, I can't confirm this issue:

CMAKE_MINIMUM_REQUIRED(VERSION 2.8 FATAL_ERROR)
PROJECT(OPTIONS C)
SET(CMAKE_VERBOSE_MAKEFILE ON)
FILE(WRITE ${CMAKE_BINARY_DIR}/main.c "int main(void){return 0;}\n")
ADD_EXECUTABLE(main main.c)

> % cmake -DCMAKE_C_FLAGS="-Os -march=native -flto 
> -freorder-blocks-and-partition" ...
> ...
> % grep CMAKE_C_FLAGS:STRING CMakeCache.txt
> CMAKE_C_FLAGS:STRING=-Os -march=native -flto -freorder-blocks-and-partition

Could you try with this exemplary project and report the results,
especially the exact command line? BTW, which shell do you use?

Regards,

Michael
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Working with MSVC10

2011-08-29 Thread Steve Casselman
Nope. It finds part of a list of required files.  For example it finds
stdarg.h but not stdlib.h. If I rename stdarg.h (in the same directory as
stdlib.h) then it can't find stdarg.h (what you would expect). 

 

I'm about ready to give up. I've check all the permissions. It's a standard
installation of Visual Studio 10 cmake finds a on file in the correct
directory but not the next file. 

 

CHECK_REQUIRED_HEADERS (  stdarg.h stddef.h ctype.h stdlib.h stdio.h
algorithm  

 functional map vector list set math.h
fcntl.h limits.h)

 

calls 

 

#required headers

MACRO(CHECK_REQUIRED_HEADERS COMPULSARY_HEADERS)

FOREACH(func ${ARGV})

   STRING(TOUPPER ${func} FUNC)

   STRING(REPLACE . _ FUNC ${FUNC})

   CHECK_INCLUDE_FILE_CXX (${func} HAVE_${FUNC})

   IF ( HAVE_${FUNC} )

   SET(_CL_HAVE_${FUNC} ${FUNC})

   ENDIF ( HAVE_${FUNC} )

   IF ( NOT HAVE_${FUNC} )

  MESSAGE ( FATAL_ERROR "${func} could not be found" )

   ENDIF ( NOT HAVE_${FUNC} )

ENDFOREACH(func ${COMPULSARY_HEADERS})

ENDMACRO(CHECK_REQUIRED_HEADERS)

 

Steve

 

 

 

From: Kitware - Lisa Avila [mailto:kitw...@kitware.com] 
Sent: Monday, August 29, 2011 5:01 AM
To: Steve Casselman
Subject: Re: Working with MSVC10

 

Hopefully you've resolved your issue - I am sure it is something simple like
a missing directory in your search path. 

Lisa



On Thu, Aug 25, 2011 at 1:21 PM, Steve Casselman  wrote:

Sorry. Just a little frustrated. 

 

Thanks

 

Steve

 

From: Kitware - Lisa Avila [mailto:kitw...@kitware.com] 
Sent: Wednesday, August 24, 2011 9:24 PM


To: Steve Casselman
Subject: Re: Working with MSVC10

 

Hello Steve,

We do have a mailing list for asking technical questions about CMake - I
provided a link to that in the previous email I sent. This email address is
not used for that purpose, and I am not a CMake expert able to answer your
question. However, I can assure you that CMake does "work" - it is tested
continuously and nightly on nearly 100 different system configurations (some
of which are Windows platforms running Visual Studio 10). 

Note that both the CMake software and the mailing list are free. And since
CMake is provided as open source, you are more than welcome to locate the
"bug" (assuming it is a bug in CMake and not in your use of it). We do have
a public Mantis bug tracker where you can enter your issue, and you can
submit a patch with the fix for inclusion in CMake. 

If you require any further information on our support, training, or
consulting options, please let me know and I'd be happy to assist you. 

Lisa

--

Lisa S. Avila
Vice President
Kitware, Inc.
28 Corporate Drive
Clifton Park, NY 12065 USA
Direct: (518) 881-4903  
Main Line: (518) 371-3971   (x504)
lisa.av...@kitware.com




On Wed, Aug 24, 2011 at 5:42 PM, Steve Casselman  wrote:

So It's funny that this stuff does not really work and you make your money
off someone finding a bug.

 

I get this

 

Looking for C++ include stdarg.h - found

Looking for C++ include stddef.h

Looking for C++ include stddef.h - found

Looking for C++ include ctype.h

Looking for C++ include ctype.h - found

Looking for C++ include stdlib.h

Looking for C++ include stdlib.h - not found

 

stdlib.h is in the same directory as everything else. 

 

Steve 

 

 

___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake

[CMake] find_package with config file always set xyz_FOUND to true?

2011-08-29 Thread Anton Deguet
Hello,

I am trying to use the following syntax:
find_package (xyz REQUIRED xyz1 xyz2) using a config file, i.e. NOT 
Findxyz.cmake.

I have a file named xyz-config.cmake that contains some code to check if the 
required components are available or not (based on how xyz was compiled).  
Ideally I would like the find_package command to set xyz_FOUND to false if:
- xyz-config.cmake file is not found (that's working by default)
OR
- xyz-config.cmake file is found BUT one or more component is missing

In my xyz-config.cmake, I tried:
- set (xyz_FOUND FALSE)  # overall package not found as required
- set (xyz2_FOUND FALSE)  # component not found, hoping find_package would 
compare this to the list xyz_FIND_COMPONENTS

But in all cases, it seems that find_package resets xyz_FOUND to 1 as long as 
the file xyz-config.cmake is found.  Is there a way to set xyz_FOUND to 0 
within xyz-config.cmake?

Anton


___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


[CMake] How can I know the target is 32-bit, or 64bit

2011-08-29 Thread Dongsheng Song
Hi,

In CMakeLists.txt, I want to know whether cmake has built-in mechanism that
I can know the target is 32-bit, or 64bit ?

Thanks,
Dongsheng
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] How can I know the target is 32-bit, or 64bit

2011-08-29 Thread Michael Wild
On Tue 30 Aug 2011 07:54:45 AM CEST, Dongsheng Song wrote:
> Hi,
> 
> In CMakeLists.txt, I want to know whether cmake has built-in mechanism that
> I can know the target is 32-bit, or 64bit ?
> 
> Thanks,
> Dongsheng

Look at the CMAKE_SIZEOF_VOID_P variable. However, normally this is 
better handled in the code through #ifdef's, because on some platforms 
(e.g. Mac) it is possible to compile for multiple architecture at once 
with a single compiler invocation. I.e., something like

gcc -arch i386 -arch ppc -arch x86_64 -arch ppc64 hello.c

will compile hello.c *four* times, first with __i386__ defined, then 
with __ppc__, followed by __x86_64__ and finally with __ppc64__. If you 
ever plan to support Mac OS X and allow your users to set the 
CMAKE_OSX_ARCHITECTURES variable to a list of architectures with 
varying bit-sex and pointer sizes, you should *not* rely on 
CMAKE_SIZEOF_VOID_P, since that will only contain the pointer size of 
the *first* architecture.

HTH

Michael
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] How can I know the target is 32-bit, or 64bit

2011-08-29 Thread J Decker
at the moment I just set an option for linux, since I don't often have
dual mode compilers... just set it in the config...

msvc has CMAKE_CL_64 which can be tested to set options.

On Mon, Aug 29, 2011 at 10:54 PM, Dongsheng Song
 wrote:
> Hi,
>
> In CMakeLists.txt, I want to know whether cmake has built-in mechanism that
> I can know the target is 32-bit, or 64bit ?
>
> Thanks,
> Dongsheng
> ___
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at 
> http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the CMake FAQ at: 
> http://www.cmake.org/Wiki/CMake_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.cmake.org/mailman/listinfo/cmake
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the CMake FAQ at: 
http://www.cmake.org/Wiki/CMake_FAQ

Follow this link to subscribe/unsubscribe:
http://www.cmake.org/mailman/listinfo/cmake