So say the local machine only has a process called 'remo' that talks to a 'remo-server' process on a remote machine fully configured with git, gradle, python, java, etc. Then I could do something like this locally:
> remo git -clone https://github.com/user/MyProject
> cd MyProject
> remo gradle build
So with 'remo' I would be able to run all manner of CLI apps that read/write to the local file system without installing anything local. The two processes would be reasonably simple:The local 'remo' CLI app does this:
- Open socket to 'remo-server' process on remote machine
- Execute remote CLI apps
- Vend file system info to 'remo-server' (LIST, HEAD, GET, PUT, DELETE)
The remote 'remo-server' process does this: - Mount a virtual file system
- Execute processes locally in vfs root dir
- Feed vfs via communication with 'remo' process
I don't think this is a great business opportunity, and it's probably a little dangerous, but it seems like it could be pretty useful. What do you think?jeff
This talk from FOSDEM describes a Linux version of "cpu" written in Go: https://fosdem.org/2022/schedule/event/plan_9_cpu_cmd/