Re: [CMake] Patch: Add NO_CACHE option to find_xxx commands

2008-07-08 Thread Yuri Timenkov


On Monday 07 July 2008 22:04:53 Enrico Scholz wrote:
 Miguel A. Figueroa-Villanueva [EMAIL PROTECTED]

 writes:
  I would like to propose the addition of a NO_CACHE keyword to the
  find_xxx commands that would treat VAR as a non-cached variable.
 
  Better wolution would be the equivalent of autoconf's './config.status
  --recheck', e.g. reinvoking cmake with empty cache but with original
  configuration options.
 
  This is not what we want. We want to keep the cache variables since
  the user might have changed it in the gui.

 No; it is exactly what you want.  Just store the original configuration
 (e.g. 'cmake' commandline or items entered into the GUI by user) somewhere
 and rerun 'cmake' on it with an empty cache.
This is possible by using initial configuration. For example, maintainer can 
prepare text file with options and paths, which should be known before  
configure steps. User after checkout (copies and) modifies it and passes to 
CMake via -C parameter.



 Enrico

 ___
 CMake mailing list
 CMake@cmake.org
 http://www.cmake.org/mailman/listinfo/cmake

___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] How to find wxWidgets with cmake under Windows?

2008-07-08 Thread Werner Smekal

Hi Miguel,


Well, Bob is setting wxWidgets_LIB_DIR in the GUI and this should then
work for him (it's the same that you are doing with the -D param). And
from the listing Bob sent me it seems that after we find
wxWidgets_LIB_DIR correctly the rest is good to go.

Have you tried it with static libs. That is,
-DwxWidgets_LIB_DIR=%WXWIN%\lib\gcc_lib ??


No, I never tried that, I need the dll for my program.



In any case, the attached file should not require you to specify the
wxWidgets_LIB_DIR anymore.


[...]

Great, so test the attached file and let me know if it works!


I tested it, and it doesn't work, although wxWidgets_LIB_DIR is set  
correctly for MinGW. But the problem is, that the module looks for the  
wrong libraries:


FIND_PATH(wxWidgets_LIB_DIR
NAMES wxpng.lib wxpngd.lib
PATHS
${WX_ROOT_DIR}/lib/${WX_LIB_DIR_PREFIX}_dll   # prefer shared
${WX_ROOT_DIR}/lib/${WX_LIB_DIR_PREFIX}_lib
DOC Path to wxWidgets libraries?
NO_DEFAULT_PATH
)

The MinGW compiler toolset adheres to the library naming convention  
for Linux and therefore the corresponding libraries are called


libwxpng.a and libwxpngd.a

So either add these names to NAME, or use another variable? I think  
any other windows compiler would name the library wxpng.lib.


HTH,
Werner

--
Dr. Werner Smekal
Institut fuer Allgemeine Physik
Technische Universitaet Wien
Wiedner Hauptstr 8-10
A-1040 Wien
Austria

email: [EMAIL PROTECTED]
web: http://www.iap.tuwien.ac.at/~smekal
phone: +43-(0)1-58801-13463 (office), +43-(0)1-58801-13469 (laboratory)
fax: +43-(0)1-58801-13499

___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] ADD_TEST comparing the command input with a file content

2008-07-08 Thread William Candillon
Hello guys,

Thank you very much for the tip! This is exactly what I needed.

Best regards,

William

On Fri, Jul 4, 2008 at 9:56 PM, Alan W. Irwin [EMAIL PROTECTED] wrote:
 On 2008-07-04 12:08+0200 William Candillon wrote:

 Hi,

 I'm using the ADD_TEST command to test the output of my executable like
 this:
 ADD_TEST(test mycommand)
 SET_TEST_PROPERTIES(test PROPERTIES PASS_REGULAR_EXPRESSION foo)

 However, I would like to compare the output with the content of a
 file. Is there a platform independent way to do that without writing a
 perl script for example ?

 You can run cmake -E compare_files file1 file2 as part of your test at
 test time (not at the normal cmake time when the tests are configured) to
 check whether file1 and file2 are the same or not.

 Alan
 __
 Alan W. Irwin

 Astronomical research affiliation with Department of Physics and Astronomy,
 University of Victoria (astrowww.phys.uvic.ca).

 Programming affiliations with the FreeEOS equation-of-state implementation
 for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
 package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of
 Linux Links project (loll.sf.net); and the Linux Brochure Project
 (lbproject.sf.net).
 __

 Linux-powered Science
 __




-- 
William Candillon
http://apdt.googlecode.com
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


[CMake] DLL runtime error R6034

2008-07-08 Thread Mehdi Rabah
Hi,

Does anyone had this error
http://msdn.microsoft.com/en-us/library/ms235560(VS.80).aspx while using a
DLL built this cmake?

My application (also built with cmake) works fine on Release mode, I get
this error only in Debug configuration. In the old project (visual studio
files only) the application seems to work in debug and release without
problems. I've looked into the compiler options but I didn't find what could
be the key difference.

Since I switched from linux to windows recently, I'm really confused. Any
ideas on how to find where the problem is coming from?

Thanks in advance,
--
Mehdi
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake

Re: [CMake] DLL runtime error R6034

2008-07-08 Thread Yuri Timenkov


On Tuesday 08 July 2008 13:38:25 Mehdi Rabah wrote:
 Hi,

 Does anyone had this error
 http://msdn.microsoft.com/en-us/library/ms235560(VS.80).aspx while using a
 DLL built this cmake?

 My application (also built with cmake) works fine on Release mode, I get
 this error only in Debug configuration. In the old project (visual studio
 files only) the application seems to work in debug and release without
 problems. I've looked into the compiler options but I didn't find what
 could be the key difference.

 Since I switched from linux to windows recently, I'm really confused. Any
 ideas on how to find where the problem is coming from?
I suppose you use VS 2005?

You should put manifest file with your DLL. For example, I have to put 
Microsoft.VC80.CRT.manifest with msvcr80.dll to run on systems without MS 
VS 2005 redistributable runtime installed.


 Thanks in advance,
 --
 Mehdi

___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


[CMake] CTest with Python unittest

2008-07-08 Thread Nicholas Yue
Hi,

  I have had a look at the FAQ for CTest but did not find any specific
answer hence I am posting here.

  I have a couple of Python projects which has unittest as part of
their code base.

  I'd like to submit their unit test run output on to a CDash server,
is there any information available to do this?

  Is this a CTest question or a CDash question ?

Regards
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] DLL runtime error R6034

2008-07-08 Thread Mehdi Rabah
On Tue, Jul 8, 2008 at 1:39 PM, Yuri Timenkov [EMAIL PROTECTED]
wrote:



 On Tuesday 08 July 2008 15:29:51 Mehdi Rabah wrote:
  Hi Yuri,
 
  CMake generate both .embed.manifest and .intermediate.manifest (but no
  .CRT.manifest). Also, I have added VS 2008 redistribuable DLLs path in my
  %PATH%.
 I mean .CRT.manifest is manifest for MSVS CRT dll. For your libraries
 manifests have different names, but they should be in same folder as your
 dll
 after installation.

 I don't know whether CMake embeds manifest or not, because we build our
 software with VS2003 :)

 I just encountered such problem once, and to solve this problem I had to
 put
 not only dll itself to application folder, but also its manifest.


Ok thanks, maybe I should use VS 2005 (I use the express version and 2003 is
not available anymore).

I have all the right manifest files in the output folder (they are generated
by cmake, I didn't have to do anything), but I still have this problem.

Any other ideas? How could I embed a manifest file in my DLL using cmake ?





 
  Thank you also for your previous answers.
  --
  Mehdi
 
  On Tue, Jul 8, 2008 at 11:47 AM, Yuri Timenkov [EMAIL PROTECTED]
 
  wrote:
   On Tuesday 08 July 2008 13:38:25 Mehdi Rabah wrote:
Hi,
   
Does anyone had this error
http://msdn.microsoft.com/en-us/library/ms235560(VS.80).aspxhttp://msdn.microsoft.com/en-us/library/ms235560%28VS.80%29.aspx
 http://msd
   n.microsoft.com/en-us/library/ms235560%28VS.80%29.aspxwhile using a
 DLL
built this cmake?
   
My application (also built with cmake) works fine on Release mode, I
get this error only in Debug configuration. In the old project
 (visual
studio files only) the application seems to work in debug and release
without problems. I've looked into the compiler options but I didn't
find what could be the key difference.
   
Since I switched from linux to windows recently, I'm really confused.
Any ideas on how to find where the problem is coming from?
  
   I suppose you use VS 2005?
  
   You should put manifest file with your DLL. For example, I have to put
   Microsoft.VC80.CRT.manifest with msvcr80.dll to run on systems
   without MS
   VS 2005 redistributable runtime installed.
  
Thanks in advance,
--
Mehdi


___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake

[CMake] CMake not searching Fortran source suffixes

2008-07-08 Thread Robert Haines

Hi all,

I wonder if anyone can help with this that I've got stuck on.

I'm building a mixed C and Fortran project that builds libraries and  
then some example applications. When building the example apps, to  
make everything more generic and reusable I do this (like in VTK with  
its Rendering, etc subdirectories I think):


set(EXAMPLE app_name)

add_executable(${EXAMPLE} ${EXAMPLE})

where the source file it needs to compile is app_name.c

So far so good as CMake finds app_name.c by searching through a list  
of known source file suffixes to complete the filename.


But when building the Fortran examples I get an error:

CMAke Error: Cannot find source file
/source/path/and/filename for target filename

Tried extensions .c .C .c++ .cc  etc and so on.

When I turn on Fortran in my project a file called  
CMakeFortranCompiler.cmake is created which lists the possible  
Fortran extensions in a variable called  
CMAKE_Fortran_SOURCE_FILE_EXTENSIONS but these it seems aren't added  
to the list CMake uses to search with.


Is this a bug? Can I add them to the search list myself? I've looked  
in vain for a variable where the search list might be stored and  
therefore appended to.


This is with 2.4.8 and 2.6.0.

Thanks for any help/pointers,
Rob

___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


[CMake] CMake support for Python

2008-07-08 Thread Nicholas Yue
Hi,

  I'd like to build and run Python unit test using CMake.

  In the keyword Project, there is support for a language tag after
the project name.

  Is there any example of using CMake to run Python unit tests?

Regards
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] DLL runtime error R6034

2008-07-08 Thread Bill Hoffman

Mehdi Rabah wrote:



Ok thanks, maybe I should use VS 2005 (I use the express version and 
2003 is not available anymore).


I have all the right manifest files in the output folder (they are 
generated by cmake, I didn't have to do anything), but I still have this 
problem.


Any other ideas? How could I embed a manifest file in my DLL using cmake ?

 
CMake automatically embeds manifest in your application.  The trick is 
the distribution of the MS run time libraries.  For that you have to use 
the .manifest files and dll's provided by MS.  There is a module 
InstallRequiredSystemLibraries.cmake that will install those libraries 
with your application.


-Bill
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


[CMake] a suggestion. [CMake 0007303]: CMake latest: Boost_LIBRARIES is not set

2008-07-08 Thread Alexander Bubnov
Hello!
I do not want make an unnecessary bug report so I decided to write this email.

I made an issue http://public.kitware.com/Bug/view.php?id=7303.
And I suppose an other issue should be originated to documentation.
Because it is not clear that Boost_LIBRARIES is used in case of
COMPONENTS.

What do you think about it?

-- 
/BR, Alexander
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


[CMake] Check if a target has been added

2008-07-08 Thread Félix C. Morency
Hi,

I posted a message yesterday and found out that the method wasn't quite
effective. If there any way of checking if a target has already been added
to the build chain (resetting it at each configure) ? I have lots of
projects depending on the same target and this produces makefile warnings
and VS. annoying popups.

Cheers,
Félix
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake

Re: [CMake] a suggestion. [CMake 0007303]: CMake latest: Boost_LIBRARIES is not set

2008-07-08 Thread Andreas Pakulat
On 08.07.08 18:29:27, Alexander Bubnov wrote:
 Hello!
 I do not want make an unnecessary bug report so I decided to write this email.
 
 I made an issue http://public.kitware.com/Bug/view.php?id=7303.
 And I suppose an other issue should be originated to documentation.
 Because it is not clear that Boost_LIBRARIES is used in case of
 COMPONENTS.

You should open a new report for that. Personally I think the
documentation is pretty clear, though maybe not very verbose on that
particular part. It reads Boost_LIBRARIES Link these to use the Boost
libraries that you specified, so obviously if you don't specify any
libraries you can expect the variable to be not set.

Andreas

-- 
You learn to write as if to someone else because NEXT YEAR YOU WILL BE
SOMEONE ELSE.
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] a suggestion. [CMake 0007303]: CMake latest: Boost_LIBRARIES is not set

2008-07-08 Thread Alexander Bubnov
I agree.

But about
2008/7/8, Andreas Pakulat [EMAIL PROTECTED]:
 On 08.07.08 18:29:27, Alexander Bubnov wrote:
   Hello!
   I do not want make an unnecessary bug report so I decided to write this 
 email.
  
   I made an issue http://public.kitware.com/Bug/view.php?id=7303.
   And I suppose an other issue should be originated to documentation.
   Because it is not clear that Boost_LIBRARIES is used in case of
   COMPONENTS.


 You should open a new report for that. Personally I think the
  documentation is pretty clear, though maybe not very verbose on that
  particular part. It reads Boost_LIBRARIES Link these to use the Boost
  libraries that you specified, so obviously if you don't specify any
  libraries you can expect the variable to be not set.

  Andreas


  --
  You learn to write as if to someone else because NEXT YEAR YOU WILL BE
  SOMEONE ELSE.
  ___
  CMake mailing list
  CMake@cmake.org
  http://www.cmake.org/mailman/listinfo/cmake



-- 
/BR, Alexander
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Newbie creating shared objects on Mac

2008-07-08 Thread Kai Sterker
On Mon, Jul 7, 2008 at 9:39 PM, Gerrick Bivins
[EMAIL PROTECTED] wrote:

 I realized I left out the fact that I'm using the swig module.
 Sorry about that.
 Does the SWIG_ADD_MODULE have a SWIG_ADD_LIBRARY equivalent?

It's the same for me, but it hasn't caused me any troubles so far.

[snip]
[ 81%] Swig source
Scanning dependencies of target _main
[ 81%] Building CXX object
src/py-wrappers/adonthell/CMakeFiles/_main.dir/py_mainPYTHON_wrap.o
[ 82%] Building CXX object
src/py-wrappers/adonthell/CMakeFiles/_main.dir/py_main.o
Linking CXX shared module _main.so
[snip]

I am not using the latest CMake, however (still happy with 2.4.7). So
unless you have a specific issue (aside from a file extension that's
atypical for a Mac shared library), I wouldn't bother.

Kai
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Newbie creating shared objects on Mac

2008-07-08 Thread Bill Hoffman

Kai Sterker wrote:

On Mon, Jul 7, 2008 at 9:39 PM, Gerrick Bivins
[EMAIL PROTECTED] wrote:


I realized I left out the fact that I'm using the swig module.
Sorry about that.
Does the SWIG_ADD_MODULE have a SWIG_ADD_LIBRARY equivalent?


It's the same for me, but it hasn't caused me any troubles so far.

[snip]
[ 81%] Swig source
Scanning dependencies of target _main
[ 81%] Building CXX object
src/py-wrappers/adonthell/CMakeFiles/_main.dir/py_mainPYTHON_wrap.o
[ 82%] Building CXX object
src/py-wrappers/adonthell/CMakeFiles/_main.dir/py_main.o
Linking CXX shared module _main.so
[snip]

I am not using the latest CMake, however (still happy with 2.4.7). So
unless you have a specific issue (aside from a file extension that's
atypical for a Mac shared library), I wouldn't bother.
Actually, this is typical for a shared module on the Mac.  Older macs 
made a distinction between linkable and loadable modules.  By convention 
loadable modules had the .so extension in many projects.


-Bill
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Newbie creating shared objects on Mac

2008-07-08 Thread Gerrick Bivins

Ok. If I understand you guys correctly, I should be good to go because .so
and .dylib now are the same in the mac OSX current architectures?

On 7/8/08 12:43 PM, Bill Hoffman [EMAIL PROTECTED] wrote:

 Kai Sterker wrote:
 On Mon, Jul 7, 2008 at 9:39 PM, Gerrick Bivins
 [EMAIL PROTECTED] wrote:
 
 I realized I left out the fact that I'm using the swig module.
 Sorry about that.
 Does the SWIG_ADD_MODULE have a SWIG_ADD_LIBRARY equivalent?
 
 It's the same for me, but it hasn't caused me any troubles so far.
 
 [snip]
 [ 81%] Swig source
 Scanning dependencies of target _main
 [ 81%] Building CXX object
 src/py-wrappers/adonthell/CMakeFiles/_main.dir/py_mainPYTHON_wrap.o
 [ 82%] Building CXX object
 src/py-wrappers/adonthell/CMakeFiles/_main.dir/py_main.o
 Linking CXX shared module _main.so
 [snip]
 
 I am not using the latest CMake, however (still happy with 2.4.7). So
 unless you have a specific issue (aside from a file extension that's
 atypical for a Mac shared library), I wouldn't bother.
 Actually, this is typical for a shared module on the Mac.  Older macs
 made a distinction between linkable and loadable modules.  By convention
 loadable modules had the .so extension in many projects.
 
 -Bill
 ___
 CMake mailing list
 CMake@cmake.org
 http://www.cmake.org/mailman/listinfo/cmake

___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Problem with Static Pattern Rules

2008-07-08 Thread Alexander Neundorf
On Tuesday 08 July 2008, Carminati Federico wrote:
 Hello,
it is possible to add a custom command to cmake with a static
 pattern rule? My problem is the following, I have a coding convention
 checker that for each c++ source file creates a violation file. The
 rule I had in my makefile was

 $(SRCS:.cxx=.viol) : %.viol : %.cxx
   bla bla

 How can I do this in cmake? The only way that seems to work is to
 create a single rule for each file with foreach. 

Yes, that's how to do it.
You can wrap this in a macro, so you get something like:

add_violation_rules(... files)

Alex
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Check if a target has been added

2008-07-08 Thread Alexander Neundorf
On Tuesday 08 July 2008, Félix C. Morency wrote:
 Hi,

 I posted a message yesterday and found out that the method wasn't quite
 effective. If there any way of checking if a target has already been added
 to the build chain (resetting it at each configure) ? I have lots of
 projects depending on the same target and this produces makefile warnings
 and VS. annoying popups.

You could try to get a target property, e.g. LOCATION from the target you are 
interested in, and see what you get. 
(this may behave differently in cmake 2.4 and 2.6)

Alex
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake


Re: [CMake] Visual Studio and Windows Mobile SDKs

2008-07-08 Thread Alexander Neundorf

On Monday 07 July 2008, you wrote:
 Hi,

 I'll submit a patch as soon as I've managed to work around some remaining
 issues. After messing around with some registry stuff, here's where I'm
 currently standing. I had a hard time to find whether VS stores the
 information on all the different SDKs in the registry or in a file.
 Concerning Windows Mobile they are stored in a config file called
 ${VSHOME}\VC\vcpackages\WCE.VCPlatform.config. Now I'm using a special XML
 parser engine derived from cmXMLParser to parse it and to get a vector of
 structs containing all the information, including the used instruction set,
 the exact SDK name, etc. I also removed the need for the additional command
 line flags. All the toolchain-based stuff moved into the generator, because
 the only important flag to set is the WINCE flag to access the platform
 specific files. Everything else can also be parsed from the SDK
 definitions. Also the compilers/assemblers are set in the generator itself,
 because basically it is just a change of executables within VS (because
 they are part of VS anyway). 

Does this actually influence the makefile generators in any way or is there 
information which could/should also be used by them ?

 Now I end up having a generator with a set of
 SDKs and, unfortunately, only one possibility to set this-PlatformName
 to one of the SDKs. 

Could this be done by setting a special cmake variable in the toolchain file 
(which is read very early) and which could then be evaluated by the 
generator ?

Alex
___
CMake mailing list
CMake@cmake.org
http://www.cmake.org/mailman/listinfo/cmake