completions: improve performance on zsh

[ZhongRuoyu]

• Have you followed the guidelines in our Contributing document?
• Have you checked to ensure there aren't other open Pull Requests for the same change?
• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you written new tests for your changes? Here's an example.
• Have you successfully run brew style with your changes locally?
• Have you successfully run brew typecheck with your changes locally?
• Have you successfully run brew tests with your changes locally?

---

By avoiding unneeded brew calls we can make zsh completions much faster.

As a rough measurement, completion for brew list used to take ~4s on my machine, but now it's instant.

[Bo98]

As a rough measurement, completion for brew list used to take ~4s on my machine, but now it's instant.

This is suspicious. brew list should already be implemented in bash and does ls under the hood. Why is the behaviour different?

[Bo98]

Try #18223. It should fix the performance for you without needing these changes.

[ZhongRuoyu]

Now that we have #18223, do we still want to keep the non-brew list changes though? Actually what I did (including removing brew list calls) was copy over what bash completions are doing (bash completions are fast). So even with the list.sh fix, completion for tap names is still slow.

Or we could consider reworking brew tap and migrating over Bash completions to calling brew commands directly too (to reduce discrepancy).

[Bo98]

Repository change is OK if we do so consistently across all completions (assuming it is shown to have actual measurable speed gains).

brew tap is a weird one as the bash completion code is actually older. In any case, if there's a noticeable speed increase it would be better for it to be implemented on command level as:

• It is shared between all completions
• It can benefit non-completion cases
• It can be unit tested in CI