Re: Please test: [CMake] CMake 2.8.4-rc1 ready for testing!
On Sunday 30 January 2011, Alexander Neundorf wrote: > On Saturday 29 January 2011, Christophe Giboudeaux wrote: > > Hi, > > > > On Tuesday 18 January 2011 21:58:24 Alexander Neundorf wrote: > > > So please, give this release candidate a good try, and let me > > > know if something fails. > > > > Quick feedback. Thanks for finally turning off the verbose makefile > > when using the codeblock generator. > > > > I didn't notice any regression but an unexpected behaviour with the > > graphviz generator. With kdepimlibs, I use > > --graphviz=kdepimlibs_graph.dot. Previously, this was generating > > only one file for the whole module. Now it seems a graph is also > > generated for *every* target: > > > > # ls kdepimlibs_graph.dot* > > kdepimlibs_graph.dot > > kdepimlibs_graph.dot.kabc_file > > kdepimlibs_graph.dot.testattendee > > kdepimlibs_graph.dot.abort > > kdepimlibs_graph.dot.kabc_file_core > > kdepimlibs_graph.dot.testblogcomment > > kdepimlibs_graph.dot.actionstatemanagertest > > kdepimlibs_graph.dot.kabcformat_binary > > kdepimlibs_graph.dot.testblogger1 > > kdepimlibs_graph.dot.addresseetest > > kdepimlibs_graph.dot.kabc_ldapkio > > etc... > > Yes. That's a feature. > The dot-file generation before was buggy and incomplete. > E.g. when running it on kdelibs it produced a file where dependencies > were missing and also targets were missing. > And if all that was in one file correctly, then the produced dot-file > is too big to be usable. At least I didn't find a way to view it in > a way that I can actually see something. > So now there is one dot-file for every target. Depending on what one wants to achieve with the graph reducing the number of edges in the graph with tred (transitive reduction filter) before processing it with dot might be an option. Regards, Ingo signature.asc Description: This is a digitally signed message part.
Re: Please test: [CMake] CMake 2.8.4-rc1 ready for testing!
On Saturday 29 January 2011, Christophe Giboudeaux wrote: > Hi, > > On Tuesday 18 January 2011 21:58:24 Alexander Neundorf wrote: > > So please, give this release candidate a good try, and let me know if > > something fails. > > Quick feedback. Thanks for finally turning off the verbose makefile when > using the codeblock generator. > > I didn't notice any regression but an unexpected behaviour with the > graphviz generator. With kdepimlibs, I use --graphviz=kdepimlibs_graph.dot. > Previously, this was generating only one file for the whole module. Now it > seems a graph is also generated for *every* target: > > # ls kdepimlibs_graph.dot* > kdepimlibs_graph.dot > kdepimlibs_graph.dot.kabc_file > kdepimlibs_graph.dot.testattendee > kdepimlibs_graph.dot.abort > kdepimlibs_graph.dot.kabc_file_core > kdepimlibs_graph.dot.testblogcomment > kdepimlibs_graph.dot.actionstatemanagertest > kdepimlibs_graph.dot.kabcformat_binary > kdepimlibs_graph.dot.testblogger1 > kdepimlibs_graph.dot.addresseetest > kdepimlibs_graph.dot.kabc_ldapkio > etc... Yes. That's a feature. The dot-file generation before was buggy and incomplete. E.g. when running it on kdelibs it produced a file where dependencies were missing and also targets were missing. And if all that was in one file correctly, then the produced dot-file is too big to be usable. At least I didn't find a way to view it in a way that I can actually see something. So now there is one dot-file for every target. You can also put a file named CMakeGraphVizOptions.cmake in CMAKE_SOURCE_DIR or CMAKE_BINARY_DIR. This can be used to set option for the dot-file generator. GRAPHVIZ_EXECUTABLES : boolean GRAPHVIZ_STATIC_LIBS: boolean GRAPHVIZ_SHARED_LIBS: boolean GRAPHVIZ_MODULE_LIBS: boolean GRAPHVIZ_TARGET_IGNORE_REGEX: regex for target names which should be skipped GRAPHVIZ_IGNORE_TARGETS: list of targets which should be skipped e.g. set(GRAPHVIZ_TARGET_IGNORE_REGEX ".+test") to exclude all tests. Yes, this is currently completely undocumented. Alex
Re: Please test: [CMake] CMake 2.8.4-rc1 ready for testing!
Hi, On Tuesday 18 January 2011 21:58:24 Alexander Neundorf wrote: > > So please, give this release candidate a good try, and let me know if > something fails. > Quick feedback. Thanks for finally turning off the verbose makefile when using the codeblock generator. I didn't notice any regression but an unexpected behaviour with the graphviz generator. With kdepimlibs, I use --graphviz=kdepimlibs_graph.dot. Previously, this was generating only one file for the whole module. Now it seems a graph is also generated for *every* target: # ls kdepimlibs_graph.dot* kdepimlibs_graph.dot kdepimlibs_graph.dot.kabc_file kdepimlibs_graph.dot.testattendee kdepimlibs_graph.dot.abort kdepimlibs_graph.dot.kabc_file_core kdepimlibs_graph.dot.testblogcomment kdepimlibs_graph.dot.actionstatemanagertest kdepimlibs_graph.dot.kabcformat_binary kdepimlibs_graph.dot.testblogger1 kdepimlibs_graph.dot.addresseetest kdepimlibs_graph.dot.kabc_ldapkio etc... Christophe signature.asc Description: This is a digitally signed message part.
Please test: [CMake] CMake 2.8.4-rc1 ready for testing!
Hi, CMake is nearing its 2.8.4 release, and a first release candidate is available. This version has a change which may affect the build of KDE, current KDE trunk, and it needs a special command line argument for KDE 4.5.0/4.5.1. Maybe you remember that in the KDE 4.5 cycle there was for a short time a build breakage of KDE trunk with CMake HEAD. This was because these versions of KDE included a copy of FindPackageHandleStandardArgs.cmake from CMake 2.8.0 or so. Now with cmake 2.8.3 FindPackageHandleStandardArgs.cmake was enhanced, and the find-modules coming with cmake made use of these new features. This broke the build, because these updated find-modules got our (KDE's) version of FindPackageHandleStandardArgs.cmake, which didn't have that feature yet. What I did back then was to add an updated version of FindPackageHandleStandardArgs.cmake to KDE, so KDE 4.5.2 works again. Still was a general problem, that a project could override files shipped with cmake, so that other cmake modules which use these files could not rely on actually getting the files they expect. So, now, with the coming 2.8.4 the behaviour of include() and find_package() has changed in this regard, and this comes with the new cmake policy CMP0017. Now, when include() or find_package() is called by a file which is part of cmake, the cmake-files installed with cmake are prefered over those with the same name in CMAKE_MODULE_PATH. This is somewhat similar to a RPATH for these files, with the "RPATH" pointing into cmake/share/Modules/. So please, give this release candidate a good try, and let me know if something fails. To build KDE 4.5.0 or 4.5.1 with it, you should have to add "-DCMAKE_POLICY_DEFAULT_CMP0017=NEW" to the cmake command. Alex -- Forwarded Message -- Subject: [CMake] CMake 2.8.4-rc1 ready for testing! Date: Thursday 13 January 2011 From: David Cole To: cm...@cmake.org I am happy to announce that CMake 2.8.4 has entered the release candidate stage! You can find the source and binaries here: http://www.cmake.org/files/v2.8/?C=M;O=D Following is the list of changes in this release. Please try this version of CMake on your projects and report any issues to the list or the bug tracker. Happy building! -Dave Changes in CMake 2.8.4-rc1 (since 2.8.3) Alex Neundorf (32): Add support for nasm assembler, patch by Peter Collingbourne (see #10069) Improve misleading comments. Add missing copyright headers We already have 2010, fix copyright year. Make FindBISON work properly with non-C locales (#11326) Add support for yasm, a nasm compatible assembler Use CMAKE_ASM_NASM_FLAGS for nasm instead of FLAGS Remove trailing whitespace and minor formatting changes for the dot-code Move the code for collecting targets and libraries into separate functions Properly insert all targets, also those which don't link to anything. Generate separate dot files for each target, and a big one with everything. Move the code for generating dot-files into separate class cmGraphVizWriter Fix #11421: FindQt3.cmake doesn't honor the REQUIRED keyword Remove trailing whitespace Don't enforce VERBOSE makefiles for the CodeBlocks generator Remove the "early alpha stage" comments about Eclipse and C::B Don't disable colors in the CodeBlocks generator and minor cleanup. Some more fixes for nasm support, from Etienne (#10069) Enable/disable generating graphs depending on the target type Use std::cout instead of fprintf Collect targets and libs on demand instead of in the ctor Exclude targets from the graphviz file based on a regex Include CMakeDetermineCompilerId in CMakeDetermineASMCompiler.cmake (#11467) Fix typos in the doc Add cache var CMAKE_ECLIPSE_MAKE_ARGUMENTS when using the Eclipse generator Add ECLIPSE_CDT4_GENERATE_SOURCE_PROJECT as a ADVANCED cache variable (#9631) Fix crash in Eclipse generator with empty project (#11616) Fix indentation in cmPolicies::ApplyPolicyVersion() Remove trailing whitespace Prefer files from CMAKE_ROOT when including from CMAKE_ROOT Improve documentation and messages for the new CMP0017 Remove usage of CMAKE_CURRENT_LIST_DIR now that we have CMP0017 Alexey Ozeritsky (5): FindBLAS works in C/C++ projects without Fortran ACML find fixes (issue 0011219) find ACML fixes fix for Fortran-only projects FindLAPACK works with C/C++ only projects (issue 0009976) Andrius Štikonas (1): Modules: Fix spelling 'becase' -> 'because'. Ben Boeckel (25): Fix parsing of cache variables without a type Use cmCacheManager to load entries from the cache Support manual cache entries Condense parsing of cache entries Use FPHSA in FindOpenGL Ignore strerror_r since CMake isn't threaded Use _P