Mantis - ISSUES

0011207: CMake 2.8 does not generate output target name properly for VS 2010 generator.

CMake 2.8 does not generate output target name properly for VS 2010 generator when there is '.' character in the file name.

Example:

add_library(My.Library STATIC ${SOURCE_FILES} ${HEADER_FILES} ${RESOURCE_FILES})

# Explicitly setting allso does NOT work
set_target_properties(My.Library PROPERTIES OUTPUT_NAME "My.Library")
set_target_properties(My.Library PROPERTIES SUFFIX ".lib")

In this scenario the generated output file is:
/OUT:"...\My.Library\Debug\My.lib" /NOLOGO

Instead of:
/OUT:"...\My.Library\Debug\My.Library.lib" /NOLOGO

Projects that depend on the My.Library target are trying to link with "My.Library.lib" resulting in a link error:

LINK : fatal error LNK1104: cannot open file '..\My.Library\Debug\My.Library.lib'

0010484: There is no way to run svn/git/hg with custom options like --username.

I have a Subversion repository with authentication.
So I have to pass "--username foo --password bar" to every /usr/bin/svn invocation.

1) The preferred way to do this is to introduce variables like CTEST_SVN_OPTIONS
2) Another way is to make cmCTestVC::CommandLineTool of type std::vector instead of std::string, thus allowing CTEST_SVN_COMMAND to be an executable with options like "svn --username foo"

0011160: FindBoost to retrieve realpath of the boost libraries

Hello,

I would like to install some of the boost components with cpack from my cmake project.
In order to create a package on Linux targets, I use for instance the following commands:

set(BoostFilesToInstall
${Boost_PYTHON_LIBRARY_RELEASE}
${Boost_SYSTEM_LIBRARY_RELEASE}
${Boost_GRAPH_LIBRARY_RELEASE}
)
install(FILES
${BoostFilesToInstall}
CONFIGURATIONS Release
DESTINATION lib
COMPONENT libraries)

However, FindBoost retrieves the links to the boost compiled libraries, rather than the files
pointed by these links. The building of a debian/mac os bundle package complain about that.
I am not 100% sure but I think cpack is trying to install the links rather than the files themselves.

So I was wondering if it would be possible to have as ${Boost_XXX_LIBRARY} libraries the result of
get_filename_component(Boost_XXX_LIBRARY ${Boost_XXX_LIBRARY} REALPATH). This way, the package would
be consistent and the dependant libraries in my project will link with boost libraries named explicitely
with their version, compiler etc.

Do you think it is a good thing to have, or the fact that using the link name is desired behaviour ?

Kind regards,
Raffi Enficiaud

0010579: CheckStructHasMember should also return the size of the type

For properly detecting e.g. large file support one should check if off_t is 64 bit and struct stat member st_size is 64 bit too. There is currently no easy way to do that so I suggest that CheckStructHasMember will additionally return the size of the member when it checks for it's presence.

0008993: FindwxWidgets.cmake always uses default toolkit

Here is a quote of the message I sent on the user list:

I am trying to use CMake to configure a very simple project that uses
wxWidgets, and I am running into a problem. The wxUSE_GUI flag is set to 0
in the Makefile generated by CMake and I therefore cannot use the GUI part
of wxWidgets.

Diving into FindwxWidgets.cmake, I see that compilation options are obtained
through wx-config. Options to wx-config are based on only four flags that I
can set or not in my CMakeLists.cmake, which are
wxWidgets_USE_{DEBUG,UNICODE,UNIVERSAL,STATIC}.

The problem is that on my Ubuntu 9.04, there are two wxWidgets configuration
in /usr/lib/wx/config. One is base-unicode-debug-2.8, the other is
gtk2-unicode-debug-2.8. The base configuration (which cannot be used for GUI
building) is the default configuration. And unfortunately, I cannot
discriminated both configurations with the four aforementionned parameters.
I am able to differentiate them only with the "--toolkit" option of
wx-config. But I cannot add this option to the wxWidgets_SELECT_OPTIONS
variable of FindwxWidgets.cmake.

What could be a work-around? I quickly browsed the wx-config script and it seems to me that the setting of 'base' as the default toolkit is more or less hard coded in it.

On request of Miguel A. Figueroa-Villanueva, here is a command line that would return the correct configuration:

wx-config --toolkit=gtk2 --cppflags

A solution could be to add a wxWidgets_USE_TOOLKIT flag that would function just like the four others wxWidgets_USE_XXX flags, and that would enable the user to select whichever toolkit s/he wants from the set of installed ones.

0011103: TIFF image fails to initialize - execution hangs. No error message.

When the TIFFReaderInternal class in itkTIFFImageIO.cxx attempts to call Initialize(), the call to TIFFNumberOfDirectories() on tif_dir.c never returns because it is caught on an infinite while-loop.

Here are the images that produce the error:

http://www.megaupload.com/?d=QFQUTWQL [^]

0011205: itkSampleToHistogramFilter does not check for HistogramSize input correctly.

In GenerateData(), the filter checks to see if the inputs, which are wrapped in SimpleDataObjectDecorator's, are NULL. However, values get initialized in the constructor. The test does not fail because it just checks to see if an exception gets thrown, not which exception is thrown.

0010610: VS2010 generator does not handle C++ files with upper case CPP extension

Create two files, File1.cpp and File2.CPP (note the case difference). With this CMakeLists.txt:

PROJECT(Bug)
ADD_EXECUTABLE(Bug File1.cpp File2.CPP)

, run the VS2010 generator. Open the solution in VS2010, do a build, and File2.CPP will not be compiled.

0011134: Allow graphs to be colored by a vtkUnsignedCharArray

vtkGraphLayoutView does not currently support coloring vertices with a vtkUnsignedCharArray. You currently must use an array with normal scalar values (e.g. a vtkIntArray) and then use vtkViewTheme's point lookup table to set up the proper mapping of integers to colors. This is awkward in some cases.

0000824: Add @ command-line option

Adding a command-line option of @ to allow parameters to be passed in via a "response file", akin to gcc and many other tools. Patch is in attachment.

0011186: FindGTK2.cmake should include new gdk-pixbuf module

Hi,

GTK2 has divorced from gdk-pixbuf.

FindGTK2.cmake should therefor now include this new library since many projects make use of it.

Patch: https://bugzilla.novell.com/show_bug.cgi?id=635439 [^]

0011206: nsis installer doesn't install a second time with components

When making a component installer, the installer doesn't work when run a second time and it skips installing files.

So for whoever maintains the NSIS cpack code....

Commit 83ec8c35933439305895b8569b2125a7e091fc57 introduced this behavior, with a comment:
- Bug fix: NSIS installers now only install/remove the minimal
number of files when re-run to update the installation (or by clicking
"Change" in Add/Remove programs)

The "Change" button isn't working for me anyway.
To make the installer work a second time, I can do this:
--- a/Source/CPack/cmCPackNSISGenerator.cxx
+++ b/Source/CPack/cmCPackNSISGenerator.cxx
@@ -627,9 +627,6 @@ CreateComponentDescription(cmCPackComponent *component,
}
componentCode += " SetOutPath \"$INSTDIR\"\n";

- componentCode += " IntCmp $" + component->Name
- + "_was_installed ${SF_SELECTED} noinstall_" + component->Name + "\n";
-
// Create the actual installation commands
if (component->IsDownloaded)
{
@@ -779,7 +776,6 @@ CreateComponentDescription(cmCPackComponent *component,
componentCode += " File /r \"${INST_DIR}\\" +
component->Name + "\\*.*\"\n";
}
- componentCode += " noinstall_" + component->Name + ":\n";
componentCode += "SectionEnd\n";

0003796: Add Project Group command

Visual Studio has "Solution Folders" for grouping Projects (i.e. build targets). Large software projects would benefit from this feature. I suppose the appropriate command would be PROJECT_GROUP, to parallel SOURCE_GROUP.

0011203: Extract surface then extract block wrong representation

Hi!

I extracted the surface of a pipe filled with some elements. Then I sliced it to calculate the wetted perimeter of the whole thing. All is fine.

I reused the extracted surface and I extracted a block from this (pipe surface without elements) and I sliced it. The integrate variable filter worked great to calculate this new perimeter (good results) but paraview keeps representing the whole perimeter, not what it actually calculated.

0009713: LOAD_CACHE command doesn't exist in CTest

I'm trying to make two different packages in a CTest script that does this:

> 1. Check out from SVN
> 2. Configure (for simulation mode)
> 3. Build
> 4. Test
> 5. Package simulation version
> 6. Copy to package archive
> 7. Reconfigure (for real mode)
> 8. Test
> 9. Package real version
> 10. Copy package to archive

It would be nice to load up the CPACK_PACKAGE_FILE_NAME from the Cache like this:

# nmake package
EXECUTE_PROCESS(WORKING_DIRECTORY "${CTEST_BINARY_DIRECTORY}"
COMMAND ${CTEST_CPACK_COMMAND}
--config ${CTEST_BINARY_DIRECTORY}/CPackConfig.cmake
RESULT_VARIABLE package_result )

# load variable
LOAD_CACHE( ${CTEST_BINARY_DIRECTORY/CMakeCache.txt
READ_WITH_PREFIX LOCAL_
CPACK_PACKAGE_FILE_NAME )

# copy file to archive
EXECUTE_PROCESS(
WORKING_DIRECTORY "${CTEST_BINARY_DIRECTORY}"
COMMAND ${CTEST_CMAKE_COMMAND}
-E copy ${LOCAL_CPACK_PACKAGE_FILE_NAME}.exe ${PACKAGE_ARCHIVE}
RESULT_VARIABLE copy_result )

0011204: FindBoost.cmake has trouble discovering libraries when both -sgd and -gd libraries are available

When specifying the Boost_USE_STATIC_LIBS to ON, the module won't find the -sgd libraries if the -gd libraries are present. I fixed this by switching the order that they're searched for, looking for the more specific library first before the more general one, which seems to have fixed the issue.

0011202: ParaView-3.9.0 crashes when applying CellDataToPointData-filter on Ensight-dataset

ParaView-3.9.0 crashes on a (large) Ensight dataset after applying the CellDataToPointData filter. The same dataset works fine with Paraview-3.6.2.

0011201: Move the coverage legend on the left

The legend of the coverage graph covers the most recent values when the coverage is above 80% preventing a good view of the progress.

It would be great to move the legend on the left part of the graph.

0011110: CMake cannot be compiled on OpenBSD 3.9

It seems, that the missing function,getpwuid_r, which causes the compilation of CMake on
OpenBSD to fail, is not available on OpenBSD 3.9.
According to the OpenBSD Manual pages, getpwuid_r was added in OpenBSD 4.4.

0010688: FindGTK2.cmake doesn't work on osx with macports

The paths for macports aren't in FindGTK2.cmake which causes it to fail. I'm attaching a patch that fixes it. It's possible that there are more modules that are affected by this, but I haven't dug through them to figure out what else needs it yet.

0010542: CMake uses usleep(), which is not available on Tru64 (OSF)

Building CMake 2.8.1 on Tru64 ("Digital UNIX V4.0G") leads to the following error:

g++ -I/tmp/cmake--2.8.1.build/Bootstrap.cmk -I/tmp/cmake-2.8.1/Source -I/tmp/cmake--2.8.1.build/Bootstrap.cmk -DKWSYS_NAMESPACE=cmsys -c /tmp/cmake-2.8.1/Source/kwsys/SystemTools.cxx -o SystemTools.o
/tmp/cmake-2.8.1/Source/kwsys/SystemTools.cxx: In static member function 'static void cmsys::SystemTools::Delay(unsigned int)':
/tmp/cmake-2.8.1/Source/kwsys/SystemTools.cxx:4135: error: 'usleep' was not declared in this scope
/tmp/cmake-2.8.1/Source/kwsys/SystemTools.cxx:4139: error: 'usleep' was not declared in this scope
gmake: *** [SystemTools.o] Error 1

I think SystemTools::Delay() is going to need that select() fallback....

0008943: NMake Makefiles does not include files with uppercase extensions

One project had FOO.CPP and FOO.H in their list of project files. While Visual Studio builds correctly built them (minor bug: they were not put into the "Source Files" and "Header Files" groups) NMake did not. I assume that they were simply not put in the list of target files. Renaming them to FOO.cpp and FOO.h fixed that.

0010388: Open Watcom Support lacking...

1) open watcom did not support 'debug all' flag passed to the linker to actually generate debug information into final DLL and EXE products.
2) open watcom support under C_FLAGS_DEBUG_INIT did not have the -d2 option to generate debug information into OBJ files.
3) flags such as CMAKE_CREATE_WIN32_EXE and property LINK_FLAGS were not passed to the linker (add )

0011189: Eclipse CDT - Unix Makefiles Generator Fails with only Eclipse, CMake installed

Hi,

Apologies if this duplicates another entry (several are close, but none close enough). I am unable to get the "Eclipse CDT - Unix Makefiles" generator to run when only Eclipse (and CMake) are installed. Below is the output when running "Configure".

------

CMake Error: CMake was unable to find a build program corresponding to "Unix Makefiles". CMAKE_MAKE_PROGRAM is not set. You probably need to select a different build tool.
CMake Error: Error required internal CMake variable not set, cmake may be not be built correctly.
Missing variable is:
CMAKE_C_COMPILER_ENV_VAR
CMake Error: Error required internal CMake variable not set, cmake may be not be built correctly.
Missing variable is:
CMAKE_C_COMPILER
CMake Error: Could not find cmake module file:C:/Users/sprinkle/Desktop/ece373/src/cmake-report-build/CMakeFiles/CMakeCCompiler.cmake
CMake Error: Error required internal CMake variable not set, cmake may be not be built correctly.
Missing variable is:
CMAKE_CXX_COMPILER_ENV_VAR
CMake Error: Error required internal CMake variable not set, cmake may be not be built correctly.
Missing variable is:
CMAKE_CXX_COMPILER
CMake Error: Could not find cmake module file:C:/Users/sprinkle/Desktop/ece373/src/cmake-report-build/CMakeFiles/CMakeCXXCompiler.cmake
CMake Error: CMAKE_C_COMPILER not set, after EnableLanguage
CMake Error: CMAKE_CXX_COMPILER not set, after EnableLanguage
Configuring incomplete, errors occurred

0011200: Document exporting to CSV from python

Document exporting to CSV from python:

From Alan:
For (dozens of times)
Open same file
Plot over line, slightly different line placement
Integrate 2d plot
Output data to spreadsheet view
Export file to .csv file.

This really is the only case I can think of for python and spreadsheet view - trying to format data such that you can export it to a csv file. But, to be able to output a filters output to a .csv file, this is the only way I know how...