Hermann von Kleist
Hermann von Kleist
> Yes, this would make sense. > However, I do not want to add Gitlab-specific code there. > It was hard enough to abstract from Travis CI in the first...
Interestingly, colcon does this: https://github.com/colcon/colcon-cmake/pull/69
Many C++ projects circumvent this problem by using `(std::min)(x, y)`. `windows.h` defines `min` as functional macro which does not get expanded when wrapping `std::min` in parens. There are other environments...
There are some alternatives to the current implementation: - `--force-color` overrides `--no-color` - Error/Warning if `--force-color` and `--no-color` are both set
I'm very sorry for the excessive abuse of the CI. I cannot run my tests locally due gtest/gmock issues on arch. :( (https://github.com/ros/catkin/issues/1030)
It looks like I need some help to get the tests working. (Non-python tests work now.) These work fine using the socket implementation, but the tests on the serial setup...
I cannot reproduce this behavior. Looking at `rosserial_client_copy_files` I wonder how this can go wrong, because the path (the one that appears to miss a trailing `/`) that is passed...
Symlinks look broken on Windows, at least they do not work out-of-the-box. Trying with redirecting includes next.
@edenhill Would you take another look at this? Thanks! I switched away from symlinks as they did not work very well on Windows. They require special treatment with Git, so...
> I think your proposal makes sense, but it has implications on a lot of derived projects that package or bundle librdkafka in various ways - so it needs to...