Skip to content

Worktrunk is a CLI for Git worktree management, designed for parallel AI agent workflows

docs(collect): reclassify AheadBehind in cache spec (#2116)

Maximilian Roose2e7934cf42b
diff --git a/src/commands/list/collect/mod.rs b/src/commands/list/collect/mod.rs
index f286cc4bc..127af4d68 100644
--- a/src/commands/list/collect/mod.rs
+++ b/src/commands/list/collect/mod.rs
@@ -132,16 +132,16 @@
//!
//! Every other task re-runs on each invocation.
//!
-//! ### Cacheable but uncached
+//! ### Already optimized (not cache candidates)
//!
-//! A few tasks take ref *names* and reduce to a SHA pair once the refs are resolved. Same
-//! SHA-pair pattern as `sha_cache`, just not wired up yet:
+//! - `AheadBehind` — batch-optimized via single `git for-each-ref %(ahead-behind:main)`
+//! (~11ms for all branches); per-branch tasks read the in-memory cache
+//! - `CommittedTreesMatch` — single `git rev-parse` resolving both tree SHAs (~1ms)
//!
-//! - `AheadBehind` — counts against the default branch
-//! - `CommittedTreesMatch` — tree equality against the integration target
-//! - `Upstream` — ahead/behind counts against the tracking branch
+//! ### Cacheable but uncached
//!
-//! Reuse `sha_cache` for any of these rather than inventing a new scheme.
+//! - `Upstream` — `ahead_behind()` against the tracking branch; same SHA-pair
+//! pattern as `sha_cache`, just not wired up yet
//!
//! ### Fundamentally uncacheable
//!