Ryan Roden-Corrent

Just had this startup, where it seemed to hang indefinitely on `ls .local/share/keybase/fs/private`. These are the logs from when I run that command: ``` 2018-12-23T17:04:06.012141-05:00 ▶ [DEBU FUSE serve.go:787] 1ac...

No objection to class-ifying them, though I also don't see a compelling reason to do so unless: 1. It simplifies some in-progress work 2. Someone just feels motivated to refactor...

Ah, cleaning up the API before making it extendeable would be nice. I can take a look unless someone else is on it already

@samyak-jain I haven't started. Go ahead, and let me know if you have any questions!

I don't think completion is part of the public API yet, so backward compatibility isn't an issue. I was imagining that we'd have models extending `CompletionModel`, and possibly categories extending...

I actually think your first one is better, as it allows for us to "pass arguments to the completion class" as @The-Compiler mentioned. Sorry for the confusion, I've actually struggled...

That makes sense to me! Returning a model feels cleaner than mutating some internal state, especially if we go with `__call__`.

> Do you think we should more strictly define what args is At a second glance I'm not sure what `args` are, so @The-Compiler might have to clarify. Are they...

> I don't quite understand how func.qute_args works 1. The `cmdutils.register` decorator [places a function](https://github.com/rcorre/qutebrowser/blob/ba1bbd555c6869215c56490c595f20259c311744/qutebrowser/api/cmdutils.py#L150-L152) maps a command's name to the decorated function in a [global object](https://github.com/rcorre/qutebrowser/blob/ba1bbd555c6869215c56490c595f20259c311744/qutebrowser/misc/objects.py#L46) 2. Each `cmdutils.argument`...

> I'm assuming you meant what I described above? Yup, that's what I was thinking > I like this approach. Likely a fix for it would also fix https://github.com/qutebrowser/qutebrowser/issues/5791 then....