On 11/27/2013 10:40 AM, Brad King wrote: > Go ahead and merge it to 'next' for testing. I will review it > there
Looking at the topic as of commit 10fd6fba, here are some comments. Why does CMAKE_PARSE_ARGUMENTS_DEFAULT_SKIP_EMPTY need to be an inherited directory property instead of a variable? Both have essentially the same scope and variables are much more common. I'd like to avoid new uses of define_property if possible. It is a legacy from the old builtin documentation system. I see this in the tests: +# run_cmake(VERSION-KEEP) # Enable when 3.0.0 is released We need to be able to test this even prior to the 3.0 release. No one is going to remember to enable that as part of the release process, and what if it is broken by then? You might be able to hack the test with set(CMAKE_MINIMUM_REQUIRED_VERSION 3.0.0) after the real cmake_minimum_required call. Stepping back for a moment: This macro is now a huge amount of CMake code to do keyword argument processing for each invocation of the calling macro or function. Accumulation of such uses adds overhead to CMake language processing. It was previously proposed in this thread to convert the implementation of the command to be in C++. IMO this is still worth investigating, if only as a follow up to this topic. Thanks, -Brad -- Powered by www.kitware.com Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html Please keep messages on-topic and check the CMake FAQ at: http://www.cmake.org/Wiki/CMake_FAQ Follow this link to subscribe/unsubscribe: http://public.kitware.com/cgi-bin/mailman/listinfo/cmake-developers