[Sprint] sprint-loop-32

[scealiontach]

Sprint plan — 2026-05-03 — sprint-loop-32

Sprint goal

Raise automated regression coverage for two command scripts that currently lack bats specs: on-change (file-watch / rerun loop) and minikube-test-environment (cluster bootstrap). Both tickets are test-only work aligned with existing patterns (tests/kind-test-environment.bats, helpers::isolate_home). The Linear shell-scripts project had only two Backlog items after filters; this sprint takes both and moves them to Todo so implementation can proceed without scope creep beyond those scripts.

Capacity note: Target selection range was 3–7 issues; only two eligible Backlog issues existed after team/project/status/label/dependency filters, so the sprint is intentionally smaller.

Selected issues

SUR-2179 — Test: add bats coverage for on-change

• Linear: SUR-2179
• Title: Test: add bats coverage for on-change
• Description summary: Add bats (or sprint regression) for bash/on-change: polling / forced-rerun loop (wait_for_change, run_command), read -t behavior, WorkingDir defaulting, log output, and option wiring. References SUR-1861-class risk (global COMMAND clobber). Suggested approach: mock find/cksum or deterministic hooks; at minimum test options:: exports and error paths for missing WorkingDir.
• Rationale: Interactive timing is fragile; shallow bats still lock option parsing, argv after --, and obvious error paths, reducing regressions before deeper integration tests.
• Definition of Done
  • New tests/on-change.bats (or agreed sprint test) exists and follows repo bats conventions (setup loads helpers.bash, helpers::isolate_home).
  • Cases cover mandatory options / exports (WatchDir, WorkingDir, WaitTime or documented defaults) and failure when required inputs are missing.
  • At least one test asserts post--- command argv is passed through to execution path without mutation (per issue intent).
  • Logging or stderr expectations are asserted where stable without flaking on timing.
  • make test / targeted bats run passes locally and in CI expectations.
  • No production behavior change unless tests prove a bug; any script fix stays minimal and covered.
• Dependencies / ordering: Can run in parallel with SUR-2184; no inter-ticket ordering. Within SUR-2179: read bash/on-change fully, mirror patterns from similar command tests, then add cases from simplest (help/errors) to argv-through---.

SUR-2184 — Test: add bats coverage for minikube-test-environment

• Linear: SUR-2184
• Title: Test: add bats coverage for minikube-test-environment
• Description summary: Add tests/minikube-test-environment.bats (none today); parity gap vs kind-test-environment coverage. Target bash/minikube-test-environment: help/subcommands, env defaults, array wiring. Use stubs on PATH for minikube/kubectl/helm, MINIKUBE_DRIVER=docker, or source-only paths similar to KIND_TEST_ENVIRONMENT_SOURCE_ONLY pattern if the script supports it.
• Rationale: Large bootstrap scripts regress on help text and argument wiring (SUR-1824-class); shallow bats catch those cheaply.
• Definition of Done
  • New tests/minikube-test-environment.bats exercises help and/or parse-only paths without requiring a real cluster.
  • Stub binaries or SOURCE_ONLY-style mode used so CI stays hermetic (document pattern in test comments only if needed).
  • Assertions cover critical defaults (e.g. driver/env) called out in the issue where observable from script output or stub invocations.
  • Parity with existing style in tests/kind-test-environment.bats (stub dir, PATH prefix, temp logs where useful).
  • make test / targeted bats passes.
  • No unrelated refactors of bash/minikube-test-environment beyond what tests require.
• Dependencies / ordering: Independent of SUR-2179. Prefer studying tests/kind-test-environment.bats first, then map analogous entry points in minikube-test-environment.

Risks + mitigations

Risk	Mitigation
Only two Backlog tickets available; no buffer if one slips	Time-box each spec; land SUR-2179 first if parallelism is limited (smaller surface).
on-change tests may flake on real timing (read -t)	Prefer deterministic stubs/mocks; avoid wall-clock assertions; test parse and dispatch paths.
minikube-test-environment may lack a clean source-only hook	Introduce minimal test seam only if necessary and covered; otherwise stub external tools entirely.
Stub PATH tests drift when script renames internals	Assert on public CLI/help strings and logged argv files, not private function names, unless matching existing kind-test pattern.
pre-commit / shellcheck on new bats	Match tests/*.bats style; run pre-commit run --all-files or bats + shellcheck as in AGENTS.md before merge.

Out of scope

• Issues in Triage, In Progress, Blocked, Done, or other projects/teams.
• Tickets labeled manual (none encountered this cycle).
• Real minikube/kind clusters or network-dependent integration in CI for this sprint.
• Packing/release (make publish), documentation tarballs, or unrelated script coverage.
• Parent epics without all children Done (no such parents in this selection).

Linear Evidence

• Linear team verified: Surinis (ce9ebfde-ff2b-4f54-90f1-c388591ca110, key SUR)
• Linear project used: shell-scripts (a43901a0-b02b-4009-aae1-a6e8903d127d)
• Query / filter used: list_issues with project: "shell-scripts", state: "Backlog" (cross-checked with team: "Surinis"); per-issue get_issue with includeRelations: true; list_issues with parentId for sub-issue checks; list_comments for SUR-2179 and SUR-2184 (no comments).
• Approx. Backlog issues reviewed: 2
• Approx. manual-labelled Backlog issues skipped: 0
• Issues skipped due to unmerged blockers: 0 []
• Issues skipped due to open-PR file overlap: 0 [] (open PR file list was empty)

Sub-issue Status

No selected issues have sub-issues in Linear. No parents were skipped for incomplete children.

Parent Issue	Sub-issue	Sub-issue Status	Eligible?
—	—	—	—

Linear State Transitions

Issue ID	Previous State	New State
SUR-2179	Backlog	Todo
SUR-2184	Backlog	Todo