rosdistro VS BinaryBuilder.jl

I have added their rosdep names (found here) to my packages.xml (see the end of this post), but even after running catkin-make and trying to run the module, the script throws a ModuleNotFound exception at the UTM package meaning that the package was never installed.

I second this. Just remove the package by removing whatever you added to the rosdistro repo when submitting/publishing the package.

The business about mapping from PyPI to system dependencies is an important one, and (having not read the entire thread) I do hope that gets some attention— it's particularly curious that it's been this long and it hasn't, given Python's often-role as a glue language.

Another example of an ecosystem maintaining mappings out to system packages is ROS and rosdep:

https://github.com/ros/rosdistro/blob/master/rosdep/base.yam...

Now it's interesting because ROS is primarily concerned with supplying a sane build-from-source story, so much of what's in the rosdep "database" is the xxxx-dev packages, but in the case of wheels, it would be more about binary dependencies, and those are auto-discoverable with ldd, shlibdeps, and the like. In Debian (and I assume other distros), the binary so packages are literally the library soname + abi versions, so if you have ldd output, you have the list of exactly what to install.

The names you use for the tag (and the other dependency tags in your package.xml) are used by rosdep to figure out what needs to be installed. The name doesn't necessarily correspond to the apt or pip package name, but most of the time it's the same. For example, for matplotlib: https://github.com/ros/rosdistro/blob/24141d9063fddd6eeca7f1db9e721fa8d600c62f/rosdep/python.yaml#L6296

Thanks for the response!

> This is difficult to answer without knowing more details.

The situation specifically is the ROS ecosystem, where metadata is managed in these package.xml files:

https://github.com/ros2/rclcpp/blob/master/rclcpp/package.xm...

The federated nature of the ecosystem has led to a culture where it's very normal to be building dozens of these at once, in the same workspace together, often from multiple repos (the repo above has four in it). So there are several build tools which automate the work of examining a source workspace and building all the packages within it in the correct topological order, respecting build_depend tags. The newest of these tools (colcon) has actually made the package.xml optional in many cases, as it can examine CMakelists, setup.py, BUILD, etc, and discover for itself what the dependencies are.

Your "distribution" of ROS is formed by listing all the packages and repos in this big file, for which there is other tooling to manage pulling dependency sources, whatever: https://github.com/ros/rosdistro/blob/master/foxy/distributi...

Anyway, so the existing ROS/nix efforts (1) seem to basically consume all of this package/distribution metadata at once and generate a giant parallel structure of nix definitions (eg https://github.com/lopsided98/nix-ros-overlay/blob/master/di...), which I fear would be completely opaque to users and any system which required everyone to leave behind these existing workflows would be an immediate non-starter.

I think the ideal scenario (and what it would look like if I built this myself based on debs) would be that you could source the "base" workspace as usual (enter the nix-shell?), and check out source, build packages as usual with colcon, the usual workspace-building tool, but there'd be an extra plugin/verb/flag for it, which would make it build each package as a nix package instead of into the usual installspace. The verb would generate the nix definitions on the fly, and probably handle the invocation and build-parallelism side of it as well.

[1]: https://github.com/acowley/ros2nix, https://github.com/lopsided98/nix-ros-overlay

Thanks for the feedback. I bumped a couple of the core devs and got the change reverted and it is now merged. It should be in the next Noetic release.

There are some efforts and the startup times are getting better with every release and there's BinaryBuilder.jl.

There is the Julia package https://github.com/JuliaPackaging/BinaryBuilder.jl which creates an environment that fakes being another, but with the correct compilers and SDKs . It's used to build all the binary dependencies

https://binarybuilder.org/. You can do it manually obviously, but this is easier.

> The main pain point is probably the lack of standard, multi-environment packaging solutions for natively compiled code.

Are you talking about something like BinaryBuilder.jl[1], which provides native binaries as julia-callable wrappers?

--

[1] https://binarybuilder.org

Julia did that for binary dependencies for a few years, with adapters for several linux platforms, homebrew, and for cross-compiled RPMs for Windows. It worked, to a degree -- less well on Windows -- but the combinatorial complexity led to many hiccups and significant maintenance effort. Each Julia package had to account for the peculiarities of each dependency across a range of dependency versions and packaging practices (linkage policies, bundling policies, naming variations, distro versions) -- and this is easier in Julia than in (C)Python because shared libraries are accessed via locally-JIT'd FFI, so there is no need to eg compile extensions for 4 different CPython ABIs (Julia also has syntactic macros which can be helpful here).

To provide a better experience for both package authors and users, as well as reducing the maintenance burden, the community has developed and migrated to a unified system called BinaryBuilder (https://binarybuilder.org) over the past 2-3 years. BinaryBuilder allows targeting all supported platforms with a single build script and also "audits" build products for common compatibility and linkage snafus (similar to some of the conda-build tooling and auditwheel). There was a nice talk at AlpineConf recently (https://alpinelinux.org/conf/) covering some of this history and detailing BinaryBuilder, although I'm not sure how to link into the video.

All that to say: it can work to an extent, but it has been tried various times before. The fact that conda and manylinux don't use system packages was not borne out of inexperience, either. The idea of "make binaries a distro packager's problem" sounds like a simplifying step, but that doesn't necessarily work out.