[PATCH 1/2] completion: more complete completion for zsh.

[David Bremner]

Vincent Breitmoser <look at my.amazin.horse> writes:

> Ah, that's a bit annoying.
>
> What we could do is just stupidly load all addresses and work from there?  On my
> machine, `notmuch address --deduplicate=address --output=address '*'` takes 0.25
> seconds for a result set of 5500 addresses.
>
Here it's 1.9s for 30k addresses

> Could also put that result set in the zsh completion cache, for slightly less
> live but immediate results on repeated calls. Is there a simple way to figure
> out on the notmuch cli if a result we received is still fresh?

notmuch count --lastmod

gives you the UUID (basically epoch) and lastmod count of the
database. That can be optimized by specifying a non-existent search
term, e.g. for me

% notmuch count --lastmod to:grendel

is quite a bit faster than

% notmuch count --lastmod