Add command completion call-budget regression tests #745

New issue

Closed

opened 2026-05-13 18:38:34 +00:00 by barrettruth · 0 comments

barrettruth commented

2026-05-13 18:38:34 +00:00

Owner

Parent tracker: #744

Scope

Add regression coverage for the command-line completion lag identified in #744.

This issue is test-only unless the tests require a tiny test helper refactor. It should not change completion behavior yet.

Problem

The current completion specs assert candidate lists, but they do not protect the performance-critical call budget. :Forge pr <Tab> can return only static-looking candidates while still synchronously invoking backend PR lookup commands.

We need tests that fail for that class of regression before changing completion behavior in follow-up issues.

Required coverage

Add completion call-budget coverage for repeated vim.fn.getcompletion() calls, preferring external-call assertions over wall-clock timing.

Cover at least:

Forge pr
Forge pr m
Forge pr re
Forge pr merge method=
release completion cold and warm cache behavior
repeated local git-backed completion for ref/target values

The key expectation for this first issue is to capture the current problematic budget clearly enough that follow-up fixes can tighten it.

Backend expectations

The tests should distinguish:

local git/file commands
backend/forge CLI commands (gh, glab, tea, shell pipelines that call them)
state/cache reads
resolver calls such as current_pr() / branch_pr()

The test model should make it possible for follow-up PRs to assert that GitHub, GitLab, and Forgejo command completion do not synchronously call backend list/detail commands.

Non-goals

Do not change completion behavior in this issue.
Do not make Ex-mode require a picker backend.
Do not add numeric PR/issue/CI subject completion.
Do not solve #744 in the same PR.

Verification

Run focused tests first:

nix develop .#ci --command just test spec/command_spec.lua spec/completion_policy_spec.lua spec/state_spec.lua

Before PR, run the full gate if practical:

nix develop .#ci --command just ci

Parent tracker: #744 ## Scope Add regression coverage for the command-line completion lag identified in #744. This issue is test-only unless the tests require a tiny test helper refactor. It should not change completion behavior yet. ## Problem The current completion specs assert candidate lists, but they do not protect the performance-critical call budget. `:Forge pr <Tab>` can return only static-looking candidates while still synchronously invoking backend PR lookup commands. We need tests that fail for that class of regression before changing completion behavior in follow-up issues. ## Required coverage Add completion call-budget coverage for repeated `vim.fn.getcompletion()` calls, preferring external-call assertions over wall-clock timing. Cover at least: - `Forge pr ` - `Forge pr m` - `Forge pr re` - `Forge pr merge method=` - release completion cold and warm cache behavior - repeated local git-backed completion for ref/target values The key expectation for this first issue is to capture the current problematic budget clearly enough that follow-up fixes can tighten it. ## Backend expectations The tests should distinguish: - local git/file commands - backend/forge CLI commands (`gh`, `glab`, `tea`, shell pipelines that call them) - state/cache reads - resolver calls such as `current_pr()` / `branch_pr()` The test model should make it possible for follow-up PRs to assert that GitHub, GitLab, and Forgejo command completion do not synchronously call backend list/detail commands. ## Non-goals - Do not change completion behavior in this issue. - Do not make Ex-mode require a picker backend. - Do not add numeric PR/issue/CI subject completion. - Do not solve #744 in the same PR. ## Verification Run focused tests first: ```sh nix develop .#ci --command just test spec/command_spec.lua spec/completion_policy_spec.lua spec/state_spec.lua ``` Before PR, run the full gate if practical: ```sh nix develop .#ci --command just ci ```

barrettruth added the

bug

label

2026-05-13 18:38:34 +00:00

barrettruth referenced this issue from a pull request that will close it,

2026-05-13 18:50:18 +00:00

Add command completion call-budget regression tests #746

barrettruth closed this issue

2026-05-13 18:50:47 +00:00

barrettruth referenced this issue

2026-05-14 14:59:23 +00:00

Tracker: make command completion non-blocking #744