date:20140508

[cmake-developers] CPack IFW generator

2014-05-08 Thread Konstantin Podsvirov

Hi CMake developers! 

I'm a newbie here. 

I like CMake and Qt. 

Qt is distributed with of his Qt Installer Framework (QtIFW). 

I decided to make friends CMake and QtIFW. 

I added CPack IFW generator and module FindIFW.cmake.

You can copy the code and test it:

git clone git://podsvirov.pro/cmake/cpackifwgenerator.git

Also available preview:

http://git.podsvirov.pro/?p=cmake/cpackifwgenerator.git;a=summary

My code is still draft, but already knows how to create good installer 
based on the components.

I want you to porobovali this and said their opinion. 

If the code you like, I'll try to support it. 

I'm sorry for my english.

--
Regards
Konstantin Podsvirov-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/cgi-bin/mailman/listinfo/cmake-developers

[cmake-developers] Weird dependency on object files generated elsewhere

2014-05-08 Thread James Bigler

I was planning on fixing a bug related to flags and separable compilation
for FindCUDA when I discovered something peculiar. I'm asking here in case
I missed something with how source files are processed.

I branched off of origin/master this morning.

I have a macro that generates some custom commands for building object
files for cuda (found in FindCUDA.cmake).

CUDA_COMPILE(CUDA_FILES test_bin.cu)
message(CUDA_FILES = ${CUDA_FILES})

ADD_EXECUTABLE(cuda_compile_example
${CUDA_FILES}
${source_files}
main.cc
external_dependency.h
)

When print out CUDA_FILES I get a full path:

CUDA_FILES =
/Users/jbigler/code/FindCUDA/build-cmake-git/src/CMakeFiles/cuda_compile.dir//./cuda_compile_generated_test_bin.cu.o

Great!

When I try to compile cuda_compile_example I get this, though:

make[2]: *** No rule to make target
`src/CMakeFiles/cuda_compile.dir/cuda_compile_generated_test_bin.cu.o',
needed by `src/cuda_compile_example'. Stop.

What I don't see in my makefiles is the build rule for the object file.

I can provide a reproducer, but it would require having CUDA installed.

This is for OSX 10.8 with Xcode 5.1 (command line utils, so Makefile).

James
--

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

Kitware offers various services to support the CMake community. For more
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/cgi-bin/mailman/listinfo/cmake-developers

Re: [cmake-developers] Weird dependency on object files generated elsewhere

2014-05-08 Thread James Bigler

This is also causing my CUDA separable compilation feature to break as well.

I'm adding object files to the source list of a target, and this works in 
certain situations, but not all.

James

From: jamesbig...@gmail.commailto:jamesbig...@gmail.com 
jamesbig...@gmail.commailto:jamesbig...@gmail.com
Date: Thursday, May 8, 2014 12:34 PM
To: cmake-developers@cmake.orgmailto:cmake-developers@cmake.org 
cmake-developers@cmake.orgmailto:cmake-developers@cmake.org
Subject: [cmake-developers] Weird dependency on object files generated elsewhere

I was planning on fixing a bug related to flags and separable compilation for 
FindCUDA when I discovered something peculiar.  I'm asking here in case I 
missed something with how source files are processed.

I branched off of origin/master this morning.

I have a macro that generates some custom commands for building object files 
for cuda (found in FindCUDA.cmake).

CUDA_COMPILE(CUDA_FILES test_bin.cuhttp://test_bin.cu)
message(CUDA_FILES = ${CUDA_FILES})

ADD_EXECUTABLE(cuda_compile_example
  ${CUDA_FILES}
  ${source_files}
  main.cc
  external_dependency.h
  )

When print out CUDA_FILES I get a full path:

CUDA_FILES = 
/Users/jbigler/code/FindCUDA/build-cmake-git/src/CMakeFiles/cuda_compile.dir//./cuda_compile_generated_test_bin.cu.o

Great!

When I try to compile cuda_compile_example I get this, though:

make[2]: *** No rule to make target 
`src/CMakeFiles/cuda_compile.dir/cuda_compile_generated_test_bin.cu.o', needed 
by `src/cuda_compile_example'.  Stop.

What I don't see in my makefiles is the build rule for the object file.

I can provide a reproducer, but it would require having CUDA installed.

This is for OSX 10.8 with Xcode 5.1 (command line utils, so Makefile).

James

---
This email message is for the sole use of the intended recipient(s) and may 
contain
confidential information.  Any unauthorized review, use, disclosure or 
distribution
is prohibited.  If you are not the intended recipient, please contact the 
sender by
reply email and destroy all copies of the original message.
---
-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/cgi-bin/mailman/listinfo/cmake-developers

Re: [cmake-developers] Weird dependency on object files generated elsewhere

2014-05-08 Thread Ben Boeckel

On Thu, May 08, 2014 at 12:34:22 -0600, James Bigler wrote:
 make[2]: *** No rule to make target
 `src/CMakeFiles/cuda_compile.dir/cuda_compile_generated_test_bin.cu.o',
 needed by `src/cuda_compile_example'.  Stop.

CMake generators will make paths relative to the build tree (I think it
is to help reduce the generated output's size).

 What I don't see in my makefiles is the build rule for the object file.

That sounds like a bug.

 I can provide a reproducer, but it would require having CUDA installed.

Please provide it :) .

Also, for the generation of the files, I don't think cuda would actually
need to be there (`ln -s /usr/bin/true ~/bin/cuda` should get you at
least that far).

--Ben
-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/cgi-bin/mailman/listinfo/cmake-developers

[CMake] Adding explicit dependencies to a target

2014-05-08 Thread Tarjei Knapstad

In our project we are generating code from CORBA IDL files by adding a
custom command that generates C++ code from the IDL files and a library
target that depends on the generated output. This works as expected,
however IDL supports include directives which are (naturally) not picked up
as dependencies by CMake. To give an example:

a.idl:

#include b.idl
#include c.idl
-

Here the file a.idl file includes b.idl and c.idl. The file a.idl is given
as input to the custom command that processes IDL files into C++, but if I
change b.idl or c.idl then the target is not rebuilt. Is it possible to add
b.idl and c.idl as explicit dependencies in some way so that my target is
rebuilt if these files change? I've made various attempts with
add_custom_target() and add_dependencies() as well as
set_source_files_properties() without any luck.

Regards,
--
Tarjei
-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

Re: [CMake] Adding explicit dependencies to a target

2014-05-08 Thread Petr Kmoch

Hi Tarjei.

add_custom_command() has a DEPENDS argument where you can list any number
of files which will act as dependencies for the custom command. So you
could extend your custom command like this:

add_custom_command(
  OUTPUT ... #as before
  COMMAND ... #as before
  MAIN_DEPENDENCY a.idl
  DEPENDS b.idl c.idl
  ... # rest as before
)

You have to keep in mind that this gets processed at configure time - if
you change the #include-s used in a.idl, you have to modify the CMakeList
and re-run CMake to get the changes picked up.

It should also be possible to use the IMPLICIT_DEPENDS argument of
add_custom_command() to request the C or C++ dependency scanner to run on
the file, moving the dependency detection to build time. But that only
works for Makefile generators.

Petr


On Thu, May 8, 2014 at 9:16 AM, Tarjei Knapstad
tarjei.knaps...@gmail.comwrote:

 In our project we are generating code from CORBA IDL files by adding a
 custom command that generates C++ code from the IDL files and a library
 target that depends on the generated output. This works as expected,
 however IDL supports include directives which are (naturally) not picked up
 as dependencies by CMake. To give an example:

 a.idl:
 
 #include b.idl
 #include c.idl
 -

 Here the file a.idl file includes b.idl and c.idl. The file a.idl is given
 as input to the custom command that processes IDL files into C++, but if I
 change b.idl or c.idl then the target is not rebuilt. Is it possible to add
 b.idl and c.idl as explicit dependencies in some way so that my target is
 rebuilt if these files change? I've made various attempts with
 add_custom_target() and add_dependencies() as well as
 set_source_files_properties() without any luck.

 Regards,
 --
 Tarjei

 --

 Powered by www.kitware.com

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

 Kitware offers various services to support the CMake community. For more
 information on each offering, please visit:

 CMake Support: http://cmake.org/cmake/help/support.html
 CMake Consulting: http://cmake.org/cmake/help/consulting.html
 CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

Re: [CMake] Adding explicit dependencies to a target

2014-05-08 Thread Tarjei Knapstad

Thanks a lot Petr, that solution works perfectly.

Regards,
--
Tarjei


On 8 May 2014 11:30, Petr Kmoch petr.km...@gmail.com wrote:

 Hi Tarjei.

 add_custom_command() has a DEPENDS argument where you can list any number
 of files which will act as dependencies for the custom command. So you
 could extend your custom command like this:

 add_custom_command(
   OUTPUT ... #as before
   COMMAND ... #as before
   MAIN_DEPENDENCY a.idl
   DEPENDS b.idl c.idl
   ... # rest as before
 )

 You have to keep in mind that this gets processed at configure time - if
 you change the #include-s used in a.idl, you have to modify the CMakeList
 and re-run CMake to get the changes picked up.

 It should also be possible to use the IMPLICIT_DEPENDS argument of
 add_custom_command() to request the C or C++ dependency scanner to run on
 the file, moving the dependency detection to build time. But that only
 works for Makefile generators.

 Petr


 On Thu, May 8, 2014 at 9:16 AM, Tarjei Knapstad tarjei.knaps...@gmail.com
  wrote:

 In our project we are generating code from CORBA IDL files by adding a
 custom command that generates C++ code from the IDL files and a library
 target that depends on the generated output. This works as expected,
 however IDL supports include directives which are (naturally) not picked up
 as dependencies by CMake. To give an example:

 a.idl:
 
 #include b.idl
 #include c.idl
 -

 Here the file a.idl file includes b.idl and c.idl. The file a.idl is
 given as input to the custom command that processes IDL files into C++, but
 if I change b.idl or c.idl then the target is not rebuilt. Is it possible
 to add b.idl and c.idl as explicit dependencies in some way so that my
 target is rebuilt if these files change? I've made various attempts with
 add_custom_target() and add_dependencies() as well as
 set_source_files_properties() without any luck.

 Regards,
 --
 Tarjei

 --

 Powered by www.kitware.com

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

 Kitware offers various services to support the CMake community. For more
 information on each offering, please visit:

 CMake Support: http://cmake.org/cmake/help/support.html
 CMake Consulting: http://cmake.org/cmake/help/consulting.html
 CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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



-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

Re: [CMake] CMake for Android projects

2014-05-08 Thread Robert Dailey

So basically I am thinking of doing this:

1. Build my NDK libraries using the android cmake toolchain file as if
I'm just building normal C++ libraries on Linux with the CMake
provided makefiles in the binary output directory.
2. Create a Java project using the 'android create project' tool and
check that into version control. The 'libs' directory will be empty.
3. When building the NDK libraries, install them to the
corresponding libs directory for that java project (I may need to
install the same library multiple times if multiple java projects
depend on it).
4. Run 'ant' by hand to build the java project.

Likewise I think I can import all projects into eclipse: Import the
cmake build directory, and then each project in the source tree for
the java side. That way I can work with it all together in the same
IDE.

Does this sound feasible or am I missing steps?

On Wed, May 7, 2014 at 10:10 PM, J Decker d3c...@gmail.com wrote:
 by putting them in a directory lib/cputype



 On Wed, May 7, 2014 at 5:19 PM, Robert Dailey rcdailey.li...@gmail.com
 wrote:

 How do I tell my android Java project where the *.so files are so it
 can package them into the APK for dynamic loading when I call
 System.loadlibrary() on the SO files?

 On Tue, May 6, 2014 at 6:59 PM, Eric Wing ewmail...@gmail.com wrote:
  So I have 4 examples I actually tried to document.
  These all use my fork/derivative of the Android-CMake toolchain, which
  I believe comes from OpenCV. (It had grown stale with later NDKs and I
  hit problems). All of these rely heavily on the external NDK module
  system (NDK_MODULE_PATH).
 
  The first and easiest (most self contained) is:
  https://bitbucket.org/ewing/hello-android-almixer
  This one invokes ant via shell scripts outside CMake.
 
  I then recently got SDL building with CMake using the same techniques.
  https://bitbucket.org/ewing/sdl_android_cleanup
  This one doesn't invoke ant because it builds only a library. There
  was actually an intermediate project I helped build after ALmixer that
  this also drew from, JavaScriptCore, and I documented my procedure
  here:
 
  https://github.com/appcelerator/hyperloop/wiki/Building-JavaScriptCore-for-Android
 
 
  Once you have SDL built, and you set your NDK_MODULE_PATH directly, I
  have an Ant example and Gradle example. The ant one is just like
  Hello-Android-ALmixer.
  https://bitbucket.org/ewing/helloandroidsdl-ant
  https://bitbucket.org/ewing/helloandroidsdl-gradle
 
  The Gradle one is a lot of hacks. Google is ditching Ant/Eclipse for
  Gradle/IntelliJ, but their NDK support is even worse in the latter
  right now.
 
  Again, once you leave the NDK, all these things live outside CMake. I
  think it would be interesting to make CMake handle all of this, but I
  can't visualize it yet. But I hope others might be able to build on my
  work, like how I've built on Android-CMake.
 
  An aside, I noticed Gradle is painfully slow. Depending on how one
  does CMake integration, I'm not sure I want it. Just to invoke it and
  for it to figure out no real work needs to be done is measured in
  seconds for me.
 
 
  Thanks,
  Eric
 
 
 
  On 5/6/14, J Decker d3c...@gmail.com wrote:
  This is a page on building to android; mostly it's about my library,
  but
  names can be replaced where required
 
  https://code.google.com/p/c-system-abstraction-component-gui/wiki/BuildingForAndroid
 
  I went wit the separate cmake projects because I end up with multiple
  android projects from the same libraries; but really it could be
  appended
  all-together
 
 
  On Tue, May 6, 2014 at 2:32 PM, Robert Dailey
  rcdailey.li...@gmail.comwrote:
 
  Well to be clear, the NDK libraries are compiled in eclipse after I
  generate eclipse makefiles in CMake. Ideally, I want the CMake script
  to also configure ant execution so that it builds java and links in
  the NDK libraries. You say just use the libraries, but I'm not sure
  what this looks like as far as CMake script is concerned.
 
  Also I think we're using the android glue stuff in our existing
  project, but I'm not really sure what android glue is or if it is
  relevant to the CMake setup.
 
  On Tue, May 6, 2014 at 2:45 PM, J Decker d3c...@gmail.com wrote:
   If you've built the sources into libs, you can just use the lib; my
  sources
   are much too complex of a tree for ndk to support to build as
   sources..
  
   The java sources are compiled at the 'ant debug/release' step...
   there
  is a
   step before that I do that is 'android.bat update project --target
   android-14 --path' which makes a few other files from the
   build.xml.
  
  
   On Tue, May 6, 2014 at 8:42 AM, Robert Dailey
   rcdailey.li...@gmail.com
   wrote:
  
   There is also the question of how to handle the NDK sources and
   integrate them into the eclipse workspace. For example, I believe
   NDK
   sources must be under the 'jni' directory, but they won't be
   structured that way in the source tree. So I'm not sure if the jni

Re: [CMake] CMake for Android projects

2014-05-08 Thread J Decker

yes basically

INSTALL( PROGRAMS  ) keeps file permissions instead of INSTALL( FILES
... ) which loses file permisisons

You don't nessiciarly have to run it by hand, can add a target that could
be built...

There is a little more work associated with release APK

---

if( CMAKE_BUILD_TYPE STREQUAL release )
  set( MORE_COMMANDS
   COMMAND jarsigner -verbose -sigalg SHA1withRSA -digestalg SHA1
-keystore ${MY_KEYSTORE} bin/${Project}-${CMAKE_BUILD_TYPE}-unsigned.apk
alias_name
   COMMAND rm -f bin/${Project}-${CMAKE_BUILD_TYPE}.apk
   COMMAND zipalign -v 4
bin/${Project}-${CMAKE_BUILD_TYPE}-unsigned.apk
bin/${Project}-${CMAKE_BUILD_TYPE}.apk
 )
endif( CMAKE_BUILD_TYPE STREQUAL release )

add_custom_target( package_apk
   COMMAND android.bat update project --target android-14 --path .
   COMMAND ant.bat ${CMAKE_BUILD_TYPE}
   ${MORE_COMMANDS}
   WORKING_DIRECTORY ${CMAKE_BINARY_DIR}/package )


add_custom_target( install_apk
COMMAND echo adb install command...
 COMMAND adb install
package/bin/${Project}-${CMAKE_BUILD_TYPE}.apk
 )

add_custom_target( uninstall_apk
COMMAND echo adb uninstall command...
 COMMAND adb uninstall org.d3x0r.${Project}
 )





On Thu, May 8, 2014 at 6:43 AM, Robert Dailey rcdailey.li...@gmail.comwrote:

 So basically I am thinking of doing this:

 1. Build my NDK libraries using the android cmake toolchain file as if
 I'm just building normal C++ libraries on Linux with the CMake
 provided makefiles in the binary output directory.
 2. Create a Java project using the 'android create project' tool and
 check that into version control. The 'libs' directory will be empty.
 3. When building the NDK libraries, install them to the
 corresponding libs directory for that java project (I may need to
 install the same library multiple times if multiple java projects
 depend on it).
 4. Run 'ant' by hand to build the java project.

 Likewise I think I can import all projects into eclipse: Import the
 cmake build directory, and then each project in the source tree for
 the java side. That way I can work with it all together in the same
 IDE.

 Does this sound feasible or am I missing steps?

 On Wed, May 7, 2014 at 10:10 PM, J Decker d3c...@gmail.com wrote:
  by putting them in a directory lib/cputype
 
 
 
  On Wed, May 7, 2014 at 5:19 PM, Robert Dailey rcdailey.li...@gmail.com
  wrote:
 
  How do I tell my android Java project where the *.so files are so it
  can package them into the APK for dynamic loading when I call
  System.loadlibrary() on the SO files?
 
  On Tue, May 6, 2014 at 6:59 PM, Eric Wing ewmail...@gmail.com wrote:
   So I have 4 examples I actually tried to document.
   These all use my fork/derivative of the Android-CMake toolchain, which
   I believe comes from OpenCV. (It had grown stale with later NDKs and I
   hit problems). All of these rely heavily on the external NDK module
   system (NDK_MODULE_PATH).
  
   The first and easiest (most self contained) is:
   https://bitbucket.org/ewing/hello-android-almixer
   This one invokes ant via shell scripts outside CMake.
  
   I then recently got SDL building with CMake using the same techniques.
   https://bitbucket.org/ewing/sdl_android_cleanup
   This one doesn't invoke ant because it builds only a library. There
   was actually an intermediate project I helped build after ALmixer that
   this also drew from, JavaScriptCore, and I documented my procedure
   here:
  
  
 https://github.com/appcelerator/hyperloop/wiki/Building-JavaScriptCore-for-Android
  
  
   Once you have SDL built, and you set your NDK_MODULE_PATH directly, I
   have an Ant example and Gradle example. The ant one is just like
   Hello-Android-ALmixer.
   https://bitbucket.org/ewing/helloandroidsdl-ant
   https://bitbucket.org/ewing/helloandroidsdl-gradle
  
   The Gradle one is a lot of hacks. Google is ditching Ant/Eclipse for
   Gradle/IntelliJ, but their NDK support is even worse in the latter
   right now.
  
   Again, once you leave the NDK, all these things live outside CMake. I
   think it would be interesting to make CMake handle all of this, but I
   can't visualize it yet. But I hope others might be able to build on my
   work, like how I've built on Android-CMake.
  
   An aside, I noticed Gradle is painfully slow. Depending on how one
   does CMake integration, I'm not sure I want it. Just to invoke it and
   for it to figure out no real work needs to be done is measured in
   seconds for me.
  
  
   Thanks,
   Eric
  
  
  
   On 5/6/14, J Decker d3c...@gmail.com wrote:
   This is a page on building to android; mostly it's about my library,
   but
   names can be replaced where required
  
  
 https://code.google.com/p/c-system-abstraction-component-gui/wiki/BuildingForAndroid
  
   I went wit the separate cmake projects because I end up with multiple
   android projects from the same libraries; but really it could be
   appended

[CMake] Reinstall Visual Studio macro

2014-05-08 Thread Scott Aron Bloom

Somehow, I corrupted the visual studio CMake macro.. I have no idea how...

However, I cant figure out how to completely remove it, and then have cmake 
re-install it

I have been googling on and off for 2-3 weeks with no success

Thanks in advance,
Scott

-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

[CMake] Cmake and libcurl

2014-05-08 Thread Dzung Nguyen

This seems to be a warning only as cmake still works.

/usr/bin/cmake: /usr/local/lib/libcurl.so.4: no version information
available (required by /usr/bin/cmake)

How to fix this error? I installed latest libcurl and reinstalled cmake.


-- 
*Dzung Nguyen*

PhD Student
Electrical Engineering and Computer Science,
Northwestern University, IL, USA
http://users.eecs.northwestern.edu/~dtn419/
-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

[CMake] undesirable Bdynamic/Bstatic link flags

2014-05-08 Thread Clinton Stimpson


I have a machine I'm trying to compile some code on, and CMake is apparently 
being smart and is inserting -Wl,-Bdynamic around some of the libraries from 
the CMAKE_CXX_IMPLICIT_LINK_LIBRARIES variable.

Is there a way to stop that?  Or is there a reason why it is doing that?

Thanks,
Clint
-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

[CMake] How to find vcvarsall.bat (e.g. at C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC)? CMAKE_LINKER?

2014-05-08 Thread Niels Dekker - address until 2018

We need to find a file named vcvarsall.bat, located in the 
installation directory of Visual C++, which may have one of the 
following path names, for example:


  C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC\
  M:\Program Files (x86)\Microsoft Visual Studio 12.0\VC\
  C:\Program Files\Microsoft Visual Studio 9.0\VC\
  etc.

How can we retrieve this directory path name within our CMakeLists.txt? 
There's no standard CMake variable like ${CMAKE_COMPILER_DIR}, right?


Looking at a CMakeCache.txt file generated for Visual C++ 12 (2013) by 
CMake 2.8.12.2, I saw exactly one variable that would help us:


  CMAKE_LINKER:FILEPATH=C:/Program Files (x86)/Microsoft Visual Studio 
12.0/VC/bin/link.exe


From there, we can easily find the path to the corresponding 
vcvarsall.bat. However, ${CMAKE_LINKER} appears to be left 
undocumented intentionally, as it appears to be an internal variable:


  0008065: CMAKE_LINKER not mentioned on --help-variables
  http://www.cmake.org/Bug/view.php?id=8065
  Closed as won't fix

Does that mean we should not use ${CMAKE_LINKER}? If so, do you have 
another suggestion on how to find vcvarsall.bat?


Kind regards, Niels
--
Niels Dekker
Scientific programmer
LKEB, Leiden University Medical Center


--

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

Re: [CMake] How to find vcvarsall.bat (e.g. at C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC)? CMAKE_LINKER?

2014-05-08 Thread David Cole


How about:

   if(EXISTS $ENV{VS110COMNTOOLS}../../VC)
  get_filename_component(VC11_DIR $ENV{VS110COMNTOOLS}../../VC 
ABSOLUTE)

   endif()
   if(EXISTS $ENV{VS120COMNTOOLS}../../VC)
  get_filename_component(VC12_DIR $ENV{VS120COMNTOOLS}../../VC 
ABSOLUTE)

   endif()
   message(STATUS VC11_DIR='${VC11_DIR}')
   message(STATUS VC12_DIR='${VC12_DIR}')


HTH,
David C.

--

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

[CMake] Custom commands always running

2014-05-08 Thread Rick McGuire

There's something I'm not understanding about custom commands.  I have need
to run a couple of custom commands during our build that generated some
output files.  They are specified like this:

# Build the rexx.img file
add_custom_command(OUTPUT rexx.img
   COMMAND ./rexximage
   DEPENDS rexximage rxapi rexxutil ${image_class_files}
${platform_rexx_img_depends}
   WORKING_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY})
add_custom_target(rexx_img ALL DEPENDS rexx.img)


It produces one output file that other targets have a dependency on
and it has dependencies on several files and several other targets.
This is working fine with everything occurring in the proper order,
but I'm a bit surprised to see this command getting run
unconditionally even if none of the command dependencies have changed.
 Not a big issue, but the fact the output file gets updated means all
of the subsequent build stages that have a dependency on the rexx.img
file also get activated.  Why is this happening and is there anything
I can do to stop this from being generated unconditionally?


Rick
-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

Re: [CMake] Custom commands always running

2014-05-08 Thread Rick McGuire

Figured out the answer on my own.  The problem was how I was specifying the
file names on the depends.  Once I added the directory to the output file
and used the directory on subsequent dependencies, this started working the
way I expected.

Rick


On Thu, May 8, 2014 at 5:39 PM, Rick McGuire object.r...@gmail.com wrote:

 There's something I'm not understanding about custom commands.  I have
 need to run a couple of custom commands during our build that generated
 some output files.  They are specified like this:

  # Build the rexx.img file
 add_custom_command(OUTPUT rexx.img
COMMAND ./rexximage
DEPENDS rexximage rxapi rexxutil ${image_class_files} 
 ${platform_rexx_img_depends}
WORKING_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY})
 add_custom_target(rexx_img ALL DEPENDS rexx.img)


 It produces one output file that other targets have a dependency on and it 
 has dependencies on several files and several other targets.  This is working 
 fine with everything occurring in the proper order, but I'm a bit surprised 
 to see this command getting run unconditionally even if none of the command 
 dependencies have changed.  Not a big issue, but the fact the output file 
 gets updated means all of the subsequent build stages that have a dependency 
 on the rexx.img file also get activated.  Why is this happening and is there 
 anything I can do to stop this from being generated unconditionally?


 Rick


-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

[CMake] Telling cmake that .map files depend on linking?

2014-05-08 Thread Charles Nicholson

Hi all-

I'm building an executable using the standard command:

add_executable(MyExe a.c b.c)

I'm adjusting the CMAKE_EXE_LINKER_FLAGS to emit a mapfile, and it works.

If I delete the mapfile, though, performing an incremental build doesn't
regenerate the mapfile.  This makes sense, since I haven't expressed to
cmake that the mapfile depends on MyExe.  In the above scenario, the
MyExe.map isn't even a target, so cmake doesn't even know it exists.

I've tried creating a custom target, but I can't create a custom command
that regenerates the mapfile, since it comes from add_executable.  I then
tried to use add_dependencies(), but that seems to only influence the build
order.

I could explicitly re-run the linker as a custom command that builds a
custom target, but that seems wasteful since linking does take a bit of
time.

It almost seems like I need some way to tell add_executable that there are
more outputs than just the executable image.  Is there a way to do this?

If anybody could point me in the right direction, I'd appreciate it!
 Thanks in advance for reading.

Best,
Charles
-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

Re: [CMake] How to find vcvarsall.bat (e.g. at C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC)? CMAKE_LINKER?

2014-05-08 Thread Niels Dekker - address until 2018

Thanks for your suggestions, J Decker and David. I find both approaches 
(using registry entry paths as hints to find_program or using 
$ENV{VS110COMNTOOLS}) quite interesting. However, I'd rather not have to 
write such code for each compiler version separately, in my CMakeLists. 
Also I have the impression that CMake already knows the directory path 
name of the selected compiler (current generator), so it would be 
helpful if CMake could tell me so directly.


We currently find vcvarsall.bat as follows:

  get_filename_component(MY_LINKER_DIR ${CMAKE_LINKER} DIRECTORY)
  find_file(MY_VCVARSALL_BAT vcvarsall.bat
${MY_LINKER_DIR}/.. ${MY_LINKER_DIR}/../..)

This appears to work fine for any Visual C++ version I have tested so 
far, including both 32 and 64 bits. However, it assumes that CMake 
generates this CMAKE_LINKER variable. Is it okay to make that assumption?


Kind regards, Niels

On 5/8/2014 11:09 PM, David Cole wrote:

How about:

if(EXISTS $ENV{VS110COMNTOOLS}../../VC)
   get_filename_component(VC11_DIR $ENV{VS110COMNTOOLS}../../VC
ABSOLUTE)
endif()
if(EXISTS $ENV{VS120COMNTOOLS}../../VC)
   get_filename_component(VC12_DIR $ENV{VS120COMNTOOLS}../../VC
ABSOLUTE)
endif()
message(STATUS VC11_DIR='${VC11_DIR}')
message(STATUS VC12_DIR='${VC12_DIR}')


--

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

[CMake] find_package() vs ExternalProject_Add()

2014-05-08 Thread Zaak Beekman

Am I correct in stating that ExternalProject_Add() cannot completely
replace find_package()?

e.g. if I point ExternalProject_Add() to download a software package
(which uses a CMake build system and exports its targets and provides
a package-config.cmake file) the only thing this does is to provide
a target to download, patch, build, test, install the software in
question, yes? It doesn't look for any project-config.cmake file, to
help setup include dirs, etc., correct?

So if a project providing a config.cmake file is downloaded, built
etc. with ExternalProject_Add one still needs to then have
find_package() to know what include directories need to be added etc.
unless I am missing something.

Clarification much appreciated.

Thanks,
Izaak Beekman
===
(301)244-9367
Princeton University Doctoral Candidate
Mechanical and Aerospace Engineering
ibeek...@princeton.edu

UMD-CP Visiting Graduate Student
Aerospace Engineering
ibeek...@umiacs.umd.edu
ibeek...@umd.edu
-- 

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

Re: [CMake] How to find vcvarsall.bat (e.g. at C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC)? CMAKE_LINKER?

2014-05-08 Thread David Cole

In CMakeCache.txt, for a Visual Studio based build where C and/or C++ 
has been enabled:


//CXX compiler
CMAKE_CXX_COMPILER:FILEPATH=C:/Program Files (x86)/Microsoft Visual 
Studio 11.0/VC/bin/cl.exe


//C compiler
CMAKE_C_COMPILER:FILEPATH=C:/Program Files (x86)/Microsoft Visual 
Studio 11.0/VC/bin/cl.exe


Those are as rock solid a basis as you'll get for a typical visual 
studio build. I think it would be wiser to depend on those than on the 
CMAKE_LINKER variable.


Microsoft has a habit of assuming things like this are internal 
implementation details, though, and they often move things around from 
version to version.


So nothing is guaranteed.

But I would think the compiler variables would be better suited if you 
don't want to rely on ENV or registry... I would point out however, 
that you don't need to rely on any cmake vars or any cmake commands 
having been run if you go with ENV or registry.



HTH,
David C.

--

Powered by www.kitware.com

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

Kitware offers various services to support the CMake community. For more 
information on each offering, please visit:

CMake Support: http://cmake.org/cmake/help/support.html
CMake Consulting: http://cmake.org/cmake/help/consulting.html
CMake Training Courses: http://cmake.org/cmake/help/training.html

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

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

41 matches

Mail list logo