Grey

I think once #380 is merged we can close this issue.

Creating subscribers on the fly will work after https://github.com/ros2-rust/ros2_rust/pull/446 is merged, but that's currently blocked on #430 getting approved. @jhdcs I think this capability will be extremely important for lifecycle...

> What's the best place to do so? [discourse](https://discourse.ros.org/t/traffic-management-interfaces-for-the-next-generation/44414/3) or [github](https://github.com/open-rmf/next_gen_prototype/issues) Whichever feels more comfortable to you! My general recommendation is to use Discourse for broad high-level conversations and then...

For visibility, I've opened https://github.com/agalbachicar/ros2_rust/pull/5 which targets this PR and addresses most of the feedback that I expect the working group will have for this PR. I recommend that others...

I think it would be reasonable to close this PR in favor of #480 The original authors for this PR are credited in the new one.

> In dynamic_sequences.rs l.200 `std::slice::from_raw_parts` is called with a null pointer which triggers the panic. This reminds me of https://github.com/ros2-rust/ros2_rust/pull/407.

> lazy_cell, which LazyLock depends on, was only made stable in Rust 1.80 If this is the only blocker, it should be possible to replace all uses of `LazyLock` with...

I also wonder if we should consider using a file extension besides `.skel`, since it sort of implies that the file will only contain a single Skeleton, which is not...

> VskParser should also follow the convention. > > io namespace would be more informative name for the namespace rather than utils I like both of those ideas :+1:

I agree with removing the need for a class from DartLoader; I just wasn't sure if we wanted to do that immediately. Getting the naming and package management to be...