4 ms·

Every time I see ROS I have nightmare flashbacks to a project i worked on using it ... I did not enjoy the cmake wrapper (catkin) -- and I bloody hate the garb

Every time I see ROS I have nightmare flashbacks to a project i worked on using it ... I did not enjoy the cmake wrapper (catkin) -- and I bloody hate the garbage fire that is the python ecosystem ... someone needs to rebuild this project in rust -- could be a million times more plug and play and also better performance ... does something like that exist?

rcxdude 6y ago
It doesn't need rebuilding in a different language (though you might as well), it just needs rebuilding competently so you have something with quality suitable for any production use, let alone anything safety critical. Sadly I've not seen anything similar in terms of an integrated ecosystem, but I do believe anyone planning on using ROS for a project of significance (i.e. something which will go beyond a prototype) is better served by creating their own version of everything ROS provides.
TheArcane 6y ago
> someone needs to rebuild this project in rust I was in agreement with you until this. C++ would be better in terms of compatibility. Rust is non-existent in the world of robotics which is mostly dominated by C++ libraries
- jefft255 6y ago
  ROS is already mostly in C++. I find it very enjoyable to use with C++. Python documentation and support is lacking, and Python 3 support is very recent. But yeah; ROS has problems and they don't have anything to do with the language. Saying "rebuild that in Rust" is missing the point. How does that address the parent's complaint which is Python support?
  - breatheoften 6y ago
    my main reason for suggesting rust was to avoid the temptation to "solve the problem of the c++ build system ecosystem" which seems to be why catkin was created -- but i think this problem is basically intractable ... Rust seems to me a reasonable language for developing the library of functionality that ROS offers -- with distribution and building and component development story all solved by the ecosystem default build mechanism ...

rcxdude 6y ago

It doesn't need rebuilding in a different language (though you might as well), it just needs rebuilding competently so you have something with quality suitable for any production use, let alone anything safety critical. Sadly I've not seen anything similar in terms of an integrated ecosystem, but I do believe anyone planning on using ROS for a project of significance (i.e. something which will go beyond a prototype) is better served by creating their own version of everything ROS provides.

TheArcane 6y ago

> someone needs to rebuild this project in rust I was in agreement with you until this. C++ would be better in terms of compatibility. Rust is non-existent in the world of robotics which is mostly dominated by C++ libraries

jefft255 6y ago
ROS is already mostly in C++. I find it very enjoyable to use with C++. Python documentation and support is lacking, and Python 3 support is very recent. But yeah; ROS has problems and they don't have anything to do with the language. Saying "rebuild that in Rust" is missing the point. How does that address the parent's complaint which is Python support?
- breatheoften 6y ago
  my main reason for suggesting rust was to avoid the temptation to "solve the problem of the c++ build system ecosystem" which seems to be why catkin was created -- but i think this problem is basically intractable ... Rust seems to me a reasonable language for developing the library of functionality that ROS offers -- with distribution and building and component development story all solved by the ecosystem default build mechanism ...