Michel Hidalgo

> I think this is doable. Though, since #131, I think the main issue is now we risk install collisions between rosidl_generate_interfaces and your custom package. @hidmic is this accurate?...

> this still puts the responsibility on the user to avoid install collisions if they are installing interfaces and custom Python code to the same module. To the extent that...

@razr thanks for reporting ! Would you be willing to submit a fix for this issue?

Hmm, to patch release an LTS distro and hope we can fast-forward those changes to `master` at some point in the future isn't ideal. But I understand the situation. What...

> So with that said, I think the right thing to do here would probably be to special-case a bytestring so that we just assign it directly without doing the...

> The PR is for eloquent because I am having issues compiling the master branch (even without the action_bridge). Any help here welcome. What kind of issues? Feature PRs against...

> maybe @launch_testing.post_shutdown_test() is a good place to add this change ? That should do. > How could we provide a clean shutdown for the process on Windows? To the...

> Just thinking out loud above - I'd be willing to review a proposal for a more crash-resistant metadata approach. Ideally one that doesn't change the end result very much....

> A slightly simpler approach might be to add a `Recorder::stop` method that can be called at will, instead of having to call the destructor for cleanup. Exposed as a...

I suspect @emersonknapp's hypothesis is spot on. E.g. there're plenty `/chatter` topic publishers in `test_security` tests that are terminated by `launch` in its current, not very nice manner.