The notes I made from my initial investigations into this.

The crate I was using is clap_complete as this is an official part of the clap project. It's lacking somewhat in documentation, though being part of clap, it will continue to be maintained.

There are a couple of different ways the shell completion can work, and they have different trade-offs:

Static completions

We generate a static completion script. This tells the shell everything it needs to know to complete all possible tedge ... commands, resulting in a very long shell script that will run completely autonomously from the tedge command. Once installed, this should always work, however:

Fish doesn't support the config key completions (although may be able to following a PR to clap)
The completions won't change if tedge changes, so if they're manually installed, they'll need to be manually updated

Dynamic completions

We generate a dynamic completion script. This basically just tells the shell to invoke tedge with some particular environment variables, and if we see those, tedge will output some information for the completion script to use and will exit. This has some slightly different characteristics:

It will always be up-to-date with the current tedge binary, so you only need to install once
If you have the completion script and try and use it with an outdated tedge binary (i.e. one that doesn't have the completions feature), you will get some error messages in your terminal when trying to use completions
It works out of the box on fish
The clap-complete feature is marked as unstable, although from a code perspective it's a very small API that we're using to enable it, and it appears to work perfectly at the moment, so I don't forsee much of a maintenance burden here

Static completions

Dynamic completions

Preview (on fish)

Generate shell completions for tedge CLI #3331

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Generate shell completions for `tedge` CLI #3331