We've been experimenting with https://github.com/withfig/autocomplete for a while and the way we're thinking of going forward is to fork the fig autocompletions into our own repo and serve our own npm module. The reasons for this approach are:
Fig has become not very active since the company behind it was acquired, with the majority of commits being automated bumps to aws specs and CLA signing. No regular accounts got anything merged in the past month despite 10 new PRs coming in:
The npm module shipped for fig is not very friendly to be consumed by a TS project, it doesn't include the TS typings for example signalling it hasn't had much external use.
To add proper support for localization which seems like it would be a very big challenge unless we actually owned the repo.
To replace emoji icons with a completion kind, it's then up to the client how those kinds map to icons.
Better support for lazy loading the specs
To get rid of things we don't want and simplifying the spec as needed (pending further investigation).
Invest in generation tooling to reduce the maintenance burden.
The Windows Terminal team has signalled interesting in using/helping to maintain it (cc @zadjii-msft, @DHowett).
We've started this exploration by forking just the code.ts and code-insiders.ts files. These will be used as our test bed for further investigating capabilities, polishing those particular specs as they've been neglected, figuring out how to proceed, etc.
We've been experimenting with https://github.com/withfig/autocomplete for a while and the way we're thinking of going forward is to fork the fig autocompletions into our own repo and serve our own npm module. The reasons for this approach are:
We've started this exploration by forking just the code.ts and code-insiders.ts files. These will be used as our test bed for further investigating capabilities, polishing those particular specs as they've been neglected, figuring out how to proceed, etc.