On 05.04.08 14:13:14, Bill Hoffman wrote: > Andreas Pakulat wrote: >> On 28.03.08 10:58:39, Bill Hoffman wrote: >>> David Thulson wrote: >>>> It would be nice if this could get into 2.6.0. Boost 1.35 is about to >>>> be released: >>>> >>>> http://lists.boost.org/boost-users/2008/03/34896.php >>>> >>>> And it at least appears that the latest version in the bug tracker >>>> will look for 1.35. Is that right? The version included in the CMake >>>> 2.6.0 Beta does not recognize Boost 1.35. >>>> >>>> David Thulson >>>> >>> Is it backwards compatible with the current one? >> >> Does it have to be 100% backwards compatible to be accepted? >> > > It would be a big plus. I really hate to break peoples builds without a > good reason. Both FindMPI and FindCurses had to be fixed for backwards > compatibility recently. Is it not 100% compatible?
Not 100% as I said in an earlier mail, it won't set Boost_LIBRARY_DIRS variable anymore if one doesn't specify a boost library to use. So yes builds that do something like find_package(Boost) link_directories(${Boost_LIBRARY_DIRS}) target_link_libraries( foo -lboost_serialization-mt ) > If not, can it be made to be? Yes, though I consider that quite a hack as its not sure that Boost_LIBRARY_DIRS really contains the libraries (it just assumes a layout where the include directory is include/ or include/boost-<version> and then replaces that with lib/ (or stage/lib on MacOSX). Though I guess a big fat warning about missing library arguments is better than completely failing to run cmake? > Does it: > > 1. find the same stuff it used to See above. > 2. have the same variables available Yes. > 3. not switch from cache to non-cache on any variables. It only switches from non-cache to cache for Boost_LIBRARY_DIR and adds variables. > If it had all of that, I would put it in right away. If not, I am more > hesitant. I'll create a new version which adds the backward compatibility stuff I mentioned above and will let you know once thats done. Andreas -- Another good night not to sleep in a eucalyptus tree. _______________________________________________ CMake mailing list CMake@cmake.org http://www.cmake.org/mailman/listinfo/cmake