> Bill Hoffman wrote: > > Clinton Stimpson wrote: > >> > >> +1 for this. I ended up having to parse the xml files myself, as I > >> couldn't find an easier way to get the info I wanted. > >> I submitted a request for this. > >> http://public.kitware.com/Bug/view.php?id=8277 > >> > >> I was trying to use ctest as a replacement for a non-portable shell > >> script that did nightly testing. > >> > > > > All of the commands have RETURN_VALUE... > > > > ctest_build(.. RETURN_VALUE val) > > ctest_update( .. RETURN_VALUE val) > > ctest_test(.. RETURN_VALUE val) > > ctest_configure(.. RETURN_VALUE val) > > > > It is in the documentation for the commands... > > > > Is that what you want? > > > > -Bill > > From testing those commands, it appears they only return error if the > attempt to do the operation failed, not if the operation itself had > errors. > I couldn't get if the build failed from ctest_build() using the VS 2005 > generator (I got a 0 for build failures and build success). I also > couldn't get the number of test failures from ctest_test(). I need > that > information so I can decide whether to package up some binaries and do > something with them. > > As for the documentation at > http://www.cmake.org/cmake/help/ctest2.6docs.html > For ctest_build() it say, "Builds the given build directory and stores > results in Build.xml." It doesn't say anything about the return value, > but in experimenting, it returned -1 if it couldn't start the build > process, and 0 for anything else. > For ctest_test() it says "Tests the given build directory and stores > results in Test.xml. The second argument is a variable that will hold > value." Valuable, or has a value? :) But what values can I get back > from it? > > Clint
First of all thank you for the quick response. (@ Clint: and thanks for submitting the request, it is a perfect match to my own request.) As for the CTEST_[...] commands, they seem to be the solution, but as Clint already mentioned, the problem is that the RETURN_VALUE of CTEST_BUILD is 0 when it was possible to trigger the build, even when at least one project build has failed. For nearly the same reason Clint mentioned I need to differ between "failed to executed building process", "build was executed, but error(s) occurred", "building was successful" and may be as a nice to have "building was successful, but warning(s) occurred". That seems not to be possible by RETURN_VALUE right now. But please correct me if I am mistaken. Kind regards, Ralf Floca ------------------------------------------------------------------ Ralf Floca DKFZ German Cancer Research Center (Deutsches Krebsforschungszentrum) Member of the Helmholtz Association E071 Research Group Software Development for Integrated Diagnostics and Therapy (SIDT) Im Neuenheimer Feld 280 D-69120 Heidelberg Telefon: +49 (6221) 42 3021 E-Mail: [EMAIL PROTECTED] Web: www.dkfz.de ------------------------------------------------------------------ _______________________________________________ CMake mailing list [email protected] http://www.cmake.org/mailman/listinfo/cmake
