go-aac/Taskfile.yml at main · tphakala/go-aac · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
# https://taskfile.dev
version: '3'

vars:
  BENCH_TIME: 2s

tasks:
  default:
    desc: Show available tasks
    cmds:
      - task --list

  build:
    desc: Build all packages
    cmds:
      - go build ./...

  test:
    desc: Run all tests (verbose)
    cmds:
      - go test -v ./...

  test:short:
    desc: Run all tests
    cmds:
      - go test ./...

  test:race:
    desc: Run tests with the race detector
    cmds:
      # Deliberately no -count=1, unlike CI's race steps, where the flag exists
      # because setup-go restores a GOCACHE from a different run and a cached ok
      # could satisfy a lane that ran nothing. Locally the cache is
      # content-addressed, so a hit means this machine really did run these
      # bytes, and the cost of the flag would be paid on every `task check`
      # rather than only on a re-run. The counter-argument is real and this is a
      # judgment call rather than a settled one: a race-detector result is not a
      # pure function of the input, so caching one does hide a flake that a
      # re-run might surface. If flake hunting starts to matter more than the
      # turnaround, add -count=1 here and to test:noasm together.
      - go test -race ./...

  test:noasm:
    desc: Run the race tests against the scalar fallback (-tags noasm)
    cmds:
      - go test -race -tags noasm ./...

  test:norace:
    desc: Run the suite without the race detector
    cmds:
      # Not redundant with test:race. Several tests deliberately shrink or skip
      # under the detector, so they only ever execute here: the edge-config soak
      # drops its in-range bitrates, TestEncoderToolWiring skips outright, and
      # pcm's TestEncodeInterleavedReuseAllocs drops its zero-allocation bound.
      # CI gained this lane for the same reason; without it locally, a green
      # `task check` says nothing about any of them.
      - go test -count=1 ./...

  bench:
    desc: Run all benchmarks
    cmds:
      - go test -bench=. -benchmem -benchtime={{.BENCH_TIME}} ./...

  vet:
    desc: Run go vet for amd64 and arm64, all three tag settings
    cmds:
      - GOARCH=amd64 go vet ./...
      - GOARCH=arm64 go vet ./...
      - GOARCH=amd64 go vet -tags noasm ./...
      - GOARCH=arm64 go vet -tags noasm ./...
      # The race-tagged files (edge_soak_race_test.go, pcm/raceflag_race_test.go)
      # are compiled by neither pass above. The race test lanes do compile them,
      # but with the compiler, not with vet, so without this they are never
      # vetted at all.
      - GOARCH=amd64 go vet -tags race ./...
      - GOARCH=arm64 go vet -tags race ./...

  lint:
    desc: Run golangci-lint for all three tag settings
    cmds:
      - golangci-lint run ./...
      # The noasm files are invisible to the default run, so a violation in
      # them only shows up here and in CI's lint matrix.
      - golangci-lint run --build-tags noasm ./...
      # Same for the race-tagged files, which no other lane lints.
      - golangci-lint run --build-tags race ./...

  tidy:
    desc: Tidy and verify go.mod
    cmds:
      - go mod tidy
      - go mod verify

  hooks:install:
    desc: Point git at the repo's .githooks directory
    cmds:
      - git config core.hooksPath .githooks

  check:
    desc: Full local gate (build, vet, lint, race tests, noasm race tests, non-race tests)
    cmds:
      - task: build
      - task: vet
      - task: lint
      - task: test:race
      # CI vets, lints and tests every tag setting, so the local gate does too;
      # a noasm-only or race-only failure is otherwise invisible until push.
      - task: test:noasm
      - task: test:norace