duel-rails 0.1.0 → 0.1.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/CHANGELOG.md +86 -1
- data/README.md +8 -4
- data/lib/duel/rails/version.rb +1 -1
- data/lib/generators/duel/install/install_generator.rb +8 -5
- data/lib/generators/duel/install/templates/duel.rb.tt +18 -0
- data/lib/generators/duel/install/templates/perf-comment.yml.tt +59 -0
- data/lib/generators/duel/install/templates/perf.yml.tt +174 -37
- metadata +2 -1
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 071eed2da6a655fb491f21479c313937875acddc148f87a9c194798492f9f998
|
|
4
|
+
data.tar.gz: fc054a813446a26d4bdc9a4d8206900bb159f29f77da64e9194b5c30e89b69ef
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 616d941adc922119cc512a5616e7a45a7ca9aa1138215112404d112f4829c45d28d0bb26452c4beb8310d9d789a62d86546b4443eab9076602a4252af710ae22
|
|
7
|
+
data.tar.gz: 48e2327b9da27b1b34f1ce6eacd4a19abf237a7ee9f893b82e5b6cb361d21c7e3644fdb709411c90988c28b3add7a8c12b62f2797346960fcb166db816a451e6
|
data/CHANGELOG.md
CHANGED
|
@@ -7,6 +7,89 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
|
|
|
7
7
|
|
|
8
8
|
## [Unreleased]
|
|
9
9
|
|
|
10
|
+
## [0.1.2] - 2026-11-02
|
|
11
|
+
|
|
12
|
+
### Added
|
|
13
|
+
|
|
14
|
+
- The `.duel.rb` that `rails generate duel:install` writes shows Duel's
|
|
15
|
+
new settings: `c.fail_on`, to fail a regression only in a pull
|
|
16
|
+
request, `c.accept_label`, a pull request label that accepts a
|
|
17
|
+
regression, and `c.push_base`, to compare a push with the last commit
|
|
18
|
+
that Duel passed.
|
|
19
|
+
- The workflow that `rails generate duel:install` writes passes a pull
|
|
20
|
+
request's labels to Duel, for `c.accept_label`, comments on a pushed
|
|
21
|
+
commit when only a pull request fails (`c.fail_on`), and with
|
|
22
|
+
`c.push_base = :last_passed`, marks a push passed with a branch of its
|
|
23
|
+
own (`duel/passed/main` for `main`).
|
|
24
|
+
- The workflow runs for a merge queue (`merge_group`), and compares its
|
|
25
|
+
group with the queue's base, so the Duel check can be required with a
|
|
26
|
+
merge queue.
|
|
27
|
+
|
|
28
|
+
### Changed
|
|
29
|
+
|
|
30
|
+
- To upgrade an app that ran `rails generate duel:install` before 0.1.2,
|
|
31
|
+
run it again, and let it overwrite `.github/workflows/perf.yml` (or
|
|
32
|
+
`perf-<its path>.yml`, for an app in a subdirectory), then carry over
|
|
33
|
+
any edits of your own. Keep your `.duel.rb`. Kept, the old workflow
|
|
34
|
+
comments on a pull request itself, and with the new one, every
|
|
35
|
+
comment comes twice.
|
|
36
|
+
- A pull request gets its comment from a second workflow that the
|
|
37
|
+
generator writes, `.github/workflows/perf-comment.yml` (or
|
|
38
|
+
`perf-<its path>-comment.yml`), which runs once the Perf workflow has
|
|
39
|
+
(`workflow_run`), with a token that can write, so a pull request from
|
|
40
|
+
a fork gets a comment too. It runs none of the pull request's code.
|
|
41
|
+
The Perf workflow no longer has a Comment job, and for an app in a
|
|
42
|
+
subdirectory, it's named for the directory, as `Perf (apps/blog)`.
|
|
43
|
+
- The workflow cancels a pull request's run when a newer push to the
|
|
44
|
+
pull request starts one. Every other run, such as a push's to `main`,
|
|
45
|
+
still runs to the end.
|
|
46
|
+
|
|
47
|
+
### Fixed
|
|
48
|
+
|
|
49
|
+
- The workflow compares a pull request with the branch that it targets,
|
|
50
|
+
not the main branch. A pull request into a stable branch, or into
|
|
51
|
+
another pull request's branch, was compared with `main`.
|
|
52
|
+
|
|
53
|
+
## [0.1.1] - 2026-10-26
|
|
54
|
+
|
|
55
|
+
### Changed
|
|
56
|
+
|
|
57
|
+
- The workflow that `rails generate duel:install` writes no longer
|
|
58
|
+
comments on every pull request. By default, it comments only when a
|
|
59
|
+
benchmark fails or Duel couldn't compare, and `c.comment` in
|
|
60
|
+
`.duel.rb` chooses otherwise (`:always` for the old behavior,
|
|
61
|
+
`:results`, or `:never`). The generated `.duel.rb` shows the setting.
|
|
62
|
+
- The workflow never edits a comment. It used to update one comment on
|
|
63
|
+
every push, and now posts a new one each time it comments, so the pull
|
|
64
|
+
request keeps every report. When a run wouldn't get a comment of its
|
|
65
|
+
own, the workflow still posts one if its last comment on the pull
|
|
66
|
+
request had another status, so after a failure the first run that
|
|
67
|
+
passes says so.
|
|
68
|
+
|
|
69
|
+
### Security
|
|
70
|
+
|
|
71
|
+
- The workflow that `rails generate duel:install` writes no longer gives
|
|
72
|
+
the job that runs a pull request's code the `pull-requests: write`
|
|
73
|
+
permission. The Duel job only reads (`contents: read`, and its
|
|
74
|
+
checkout keeps no credentials), and keeps the report as an artifact
|
|
75
|
+
for a day. A Comment job of its own, which checks nothing out and runs
|
|
76
|
+
none of the app's code, downloads the report and comments with it, for
|
|
77
|
+
a pull request from the same repository.
|
|
78
|
+
|
|
79
|
+
### Fixed
|
|
80
|
+
|
|
81
|
+
- With `c.ruby = :version_file`, the workflow installs the bundle for
|
|
82
|
+
the baseline's Ruby too, when it differs from the candidate's. The
|
|
83
|
+
baseline's workers load that bundle, and gems installed for one Ruby
|
|
84
|
+
version don't load on another, so a branch that changed the Ruby's
|
|
85
|
+
minor version failed before it compared anything.
|
|
86
|
+
|
|
87
|
+
- With a `.ruby-version`, the generated workflow installs the baseline's
|
|
88
|
+
Ruby for every event that compares, not only for a pull request: the
|
|
89
|
+
commit before a push, the last release on a schedule, and where HEAD
|
|
90
|
+
left the main branch for a pull request and a manual run, when its
|
|
91
|
+
`.ruby-version` names another Ruby than HEAD's.
|
|
92
|
+
|
|
10
93
|
## [0.1.0] - 2026-10-24
|
|
11
94
|
|
|
12
95
|
The first release. duel-rails is a Duel plugin for Rails apps, open
|
|
@@ -95,5 +178,7 @@ Rails 7.2 or later, and Ruby 3.4 or later.
|
|
|
95
178
|
CI.
|
|
96
179
|
- RBS signatures in sig/, checked with Steep.
|
|
97
180
|
|
|
98
|
-
[Unreleased]: https://github.com/sferik/duel-ruby/compare/duel-rails-v0.1.
|
|
181
|
+
[Unreleased]: https://github.com/sferik/duel-ruby/compare/duel-rails-v0.1.2...HEAD
|
|
182
|
+
[0.1.2]: https://github.com/sferik/duel-ruby/compare/duel-rails-v0.1.1...duel-rails-v0.1.2
|
|
183
|
+
[0.1.1]: https://github.com/sferik/duel-ruby/compare/duel-rails-v0.1.0...duel-rails-v0.1.1
|
|
99
184
|
[0.1.0]: https://github.com/sferik/duel-ruby/releases/tag/duel-rails-v0.1.0
|
data/README.md
CHANGED
|
@@ -68,7 +68,8 @@ production does, with everything of its own:
|
|
|
68
68
|
| `config/environments/benchmark.rb` | The environment that each version boots in, based on production. |
|
|
69
69
|
| `duel/requests_duel.rb` | A benchmark of each of your app's first 10 routes that a GET reaches without parameters. |
|
|
70
70
|
| `duel/boot_duel.rb` | A benchmark of your app's boot. See [Benchmark your app's boot](#benchmark-your-apps-boot). |
|
|
71
|
-
| `.github/workflows/perf.yml` | A GitHub Actions workflow that runs Duel on pull requests, on pushes to `main`, and nightly against your last release, and
|
|
71
|
+
| `.github/workflows/perf.yml` | A GitHub Actions workflow that runs Duel on pull requests, on pushes to `main`, and nightly against your last release, and has the second workflow comment on a pull request with Duel's report when a benchmark fails. It never edits a comment, and after a failure, posts one more when a run passes. To comment on every run, or never, set `c.comment` in `.duel.rb`: see [Choose when to comment on a pull request](https://github.com/sferik/duel-ruby/blob/main/docs/configuration.md#choose-when-to-comment-on-a-pull-request) and [Report on a pull request](https://github.com/sferik/duel-ruby#report-on-a-pull-request). For an app in a subdirectory of its repository, the workflow is at the repository's root. See [An app in a monorepo](#an-app-in-a-monorepo). |
|
|
72
|
+
| `.github/workflows/perf-comment.yml` | A second workflow, which comments on a pull request with the report once the first has run, with a token that can write, so a pull request from a fork gets a comment too. It runs none of the pull request's code. See [Comment on a pull request](https://github.com/sferik/duel-ruby/blob/main/docs/ci.md#comment-on-a-pull-request). |
|
|
72
73
|
|
|
73
74
|
The generator also adds `gem "stackprof", platforms: :mri` to your
|
|
74
75
|
`Gemfile`'s `development` group, unless it has StackProf already, so
|
|
@@ -326,8 +327,9 @@ Each Ruby must be installed where Duel finds it: with rbenv, chruby,
|
|
|
326
327
|
mise, or asdf, or as the Ruby that runs Duel. For more information, see
|
|
327
328
|
Duel's [Run each version on its own Ruby](https://github.com/sferik/duel-ruby/blob/main/docs/configuration.md#run-each-version-on-its-own-ruby).
|
|
328
329
|
On GitHub Actions, the generated workflow installs your app's Ruby, and,
|
|
329
|
-
|
|
330
|
-
|
|
330
|
+
when the baseline's `.ruby-version` names another Ruby, that one too, so
|
|
331
|
+
a pull request, a push, or a nightly run against your last release
|
|
332
|
+
works there when your app upgrades Ruby.
|
|
331
333
|
|
|
332
334
|
To run both versions on the Ruby that runs Duel, remove
|
|
333
335
|
`c.ruby = :version_file`.
|
|
@@ -428,7 +430,9 @@ end
|
|
|
428
430
|
```
|
|
429
431
|
|
|
430
432
|
duel-rails also tells Duel to compare a change to `app/`, `config/`,
|
|
431
|
-
`db/`, `lib/`, `Gemfile`, `Gemfile.lock`, or `.ruby-version
|
|
433
|
+
`db/`, `lib/`, `Gemfile`, `Gemfile.lock`, or `.ruby-version`, as well
|
|
434
|
+
as one to your benchmarks (`duel/`) or `.duel.rb`, which Duel always
|
|
435
|
+
watches.
|
|
432
436
|
|
|
433
437
|
## An app in a monorepo
|
|
434
438
|
|
data/lib/duel/rails/version.rb
CHANGED
|
@@ -43,13 +43,16 @@ module Duel
|
|
|
43
43
|
|
|
44
44
|
def create_boot_benchmark = template("boot_duel.rb.tt", "duel/boot_duel.rb")
|
|
45
45
|
|
|
46
|
-
# The
|
|
47
|
-
#
|
|
48
|
-
#
|
|
46
|
+
# The workflows, at the repository's root: perf.yml, and
|
|
47
|
+
# perf-comment.yml, which comments on a pull request with the
|
|
48
|
+
# report, or, for an app in a subdirectory, perf-<its path>.yml
|
|
49
|
+
# (perf-apps-blog.yml) and perf-<its path>-comment.yml, so that each
|
|
50
|
+
# of a monorepo's apps can have its own.
|
|
49
51
|
def create_workflow
|
|
50
52
|
dir = working_directory
|
|
51
|
-
name = dir ? "perf-#{dir.tr("/", "-")}
|
|
52
|
-
template("perf.yml.tt", File.join(repository, ".github", "workflows", name))
|
|
53
|
+
name = dir ? "perf-#{dir.tr("/", "-")}" : "perf"
|
|
54
|
+
template("perf.yml.tt", File.join(repository, ".github", "workflows", "#{name}.yml"))
|
|
55
|
+
template("perf-comment.yml.tt", File.join(repository, ".github", "workflows", "#{name}-comment.yml"))
|
|
53
56
|
end
|
|
54
57
|
|
|
55
58
|
# The plugin that gives each worker its database's server.
|
|
@@ -20,4 +20,22 @@ Duel.configure do |c|
|
|
|
20
20
|
# fixtures: :all, or seeds: true.
|
|
21
21
|
c.use :rails, database: <%= database ? ":#{database}" : "nil" %>
|
|
22
22
|
<% end -%>
|
|
23
|
+
|
|
24
|
+
# When the workflow comments on a pull request with the report: :always,
|
|
25
|
+
# :results (passed or failed), :failures (failed, or couldn't compare),
|
|
26
|
+
# or :never.
|
|
27
|
+
# c.comment = :failures
|
|
28
|
+
|
|
29
|
+
# Which runs a regression fails: :every_run, or :pull_requests, for a
|
|
30
|
+
# project where people push to main directly (on GitHub Actions, another
|
|
31
|
+
# run warns and passes, and a pushed commit gets the comment instead).
|
|
32
|
+
# c.fail_on = :every_run
|
|
33
|
+
|
|
34
|
+
# A label that accepts a regression in a pull request (nil for none).
|
|
35
|
+
# c.accept_label = "duel: accept"
|
|
36
|
+
|
|
37
|
+
# What the workflow compares a push with: :before (the commit before it),
|
|
38
|
+
# or :last_passed (the last commit Duel passed, which a branch,
|
|
39
|
+
# duel/passed/main, marks), so small slowdowns add up until they fail.
|
|
40
|
+
# c.push_base = :before
|
|
23
41
|
end
|
|
@@ -0,0 +1,59 @@
|
|
|
1
|
+
name: Perf comment<%= " (#{working_directory})" if working_directory %>
|
|
2
|
+
|
|
3
|
+
# Duel's report, as a new comment on the pull request, when Duel says to:
|
|
4
|
+
# always (post), or only when its last comment's status differs from the
|
|
5
|
+
# run's (change), so the latest comment is never stale. It never edits or
|
|
6
|
+
# deletes a comment. It runs once the Perf workflow has, with this
|
|
7
|
+
# repository's own workflow and a token that can write, so a pull request
|
|
8
|
+
# from a fork gets a comment too. It checks nothing out and runs none of
|
|
9
|
+
# the pull request's code, and of what the Duel job kept (where that code
|
|
10
|
+
# ran), it trusts only what that job could have said: the pull request's
|
|
11
|
+
# number, if it names the pull request the run compared, and the comment
|
|
12
|
+
# and the status, if they're ones Duel says. (The workflow must be on the
|
|
13
|
+
# default branch before it runs at all.)
|
|
14
|
+
on:
|
|
15
|
+
workflow_run:
|
|
16
|
+
workflows: ["Perf<%= " (#{working_directory})" if working_directory %>"]
|
|
17
|
+
types: [completed]
|
|
18
|
+
|
|
19
|
+
permissions: {}
|
|
20
|
+
|
|
21
|
+
jobs:
|
|
22
|
+
comment:
|
|
23
|
+
name: Comment
|
|
24
|
+
if: ${{ github.event.workflow_run.event == 'pull_request' && github.event.workflow_run.conclusion != 'cancelled' }}
|
|
25
|
+
runs-on: ubuntu-latest
|
|
26
|
+
permissions:
|
|
27
|
+
actions: read # to get the report from the Perf run
|
|
28
|
+
pull-requests: write # to comment on the pull request with the report
|
|
29
|
+
steps:
|
|
30
|
+
- name: Get the report
|
|
31
|
+
id: report
|
|
32
|
+
continue-on-error: true # there's none when the Duel job failed before comparing
|
|
33
|
+
uses: actions/download-artifact@v8
|
|
34
|
+
with:
|
|
35
|
+
name: duel-report
|
|
36
|
+
path: ${{ runner.temp }}/report
|
|
37
|
+
run-id: ${{ github.event.workflow_run.id }}
|
|
38
|
+
github-token: ${{ github.token }}
|
|
39
|
+
- name: Comment on the pull request
|
|
40
|
+
if: steps.report.outcome == 'success'
|
|
41
|
+
env:
|
|
42
|
+
GH_TOKEN: ${{ github.token }}
|
|
43
|
+
HEAD_SHA: ${{ github.event.workflow_run.head_sha }}
|
|
44
|
+
run: |
|
|
45
|
+
said() { grep -xE "$1=($2)" "$RUNNER_TEMP/report/comment.env" | tail -n 1 | cut -d= -f2; }
|
|
46
|
+
PR=$(said pr '[0-9]+')
|
|
47
|
+
COMMENT=$(said comment 'post|change')
|
|
48
|
+
STATUS=$(said status 'passed|failed|neutral')
|
|
49
|
+
[ -n "$PR" ] && [ -n "$COMMENT" ] && [ -n "$STATUS" ] || exit 0
|
|
50
|
+
[ "$(gh api "repos/$GITHUB_REPOSITORY/pulls/$PR" --jq .head.sha)" = "$HEAD_SHA" ] || exit 0
|
|
51
|
+
marker="<!-- duel<%= " #{working_directory}" if working_directory %>"
|
|
52
|
+
# The first line of the last comment Duel posted here: its marker,
|
|
53
|
+
# with its status (none in Duel 0.1.0's, which counts as another).
|
|
54
|
+
last=$(gh api --paginate "repos/$GITHUB_REPOSITORY/issues/$PR/comments" \
|
|
55
|
+
--jq ".[] | select(.user.login == \"github-actions[bot]\") | .body | split(\"\n\")[0] | rtrimstr(\"\r\") | select(startswith(\"$marker status=\") or . == \"$marker -->\")" | tail -n 1)
|
|
56
|
+
if [ "$COMMENT" = post ] || { [ -n "$last" ] && [ "$last" != "$marker status=$STATUS -->" ]; }; then
|
|
57
|
+
{ echo "$marker status=$STATUS -->"; cat "$RUNNER_TEMP/report/duel.md"; } > "$RUNNER_TEMP/comment.md"
|
|
58
|
+
gh api --method POST "repos/$GITHUB_REPOSITORY/issues/$PR/comments" -F body=@"$RUNNER_TEMP/comment.md" > /dev/null
|
|
59
|
+
fi
|
|
@@ -1,21 +1,40 @@
|
|
|
1
|
-
name: Perf
|
|
1
|
+
name: Perf<%= " (#{working_directory})" if working_directory %>
|
|
2
2
|
|
|
3
3
|
on:
|
|
4
4
|
pull_request:
|
|
5
|
+
merge_group: # (a merge queue's)
|
|
5
6
|
push:
|
|
6
|
-
branches: [main]
|
|
7
|
+
branches: [main] # (and any other branch to guard, a stable one, say)
|
|
7
8
|
schedule:
|
|
8
9
|
- cron: "0 4 * * *"
|
|
9
10
|
workflow_dispatch:
|
|
10
11
|
|
|
12
|
+
# A pull request's newer push cancels its run, if that's still going, as
|
|
13
|
+
# only its newest commit's result counts. Every other run (a push's, say)
|
|
14
|
+
# runs to the end, so that each push is compared.
|
|
15
|
+
concurrency:
|
|
16
|
+
group: ${{ github.workflow }}-${{ github.event_name == 'pull_request' && github.event.pull_request.number || github.run_id }}
|
|
17
|
+
cancel-in-progress: ${{ github.event_name == 'pull_request' }}
|
|
18
|
+
|
|
19
|
+
# The Duel job runs the branch's code (its benchmarks too), so it only
|
|
20
|
+
# reads. Commenting on a commit is the Comment on the commit job's, which
|
|
21
|
+
# runs none of it, and commenting on a pull request is the Perf comment
|
|
22
|
+
# workflow's.
|
|
11
23
|
permissions:
|
|
12
24
|
contents: read
|
|
13
|
-
pull-requests:
|
|
25
|
+
pull-requests: read # for the pull request's labels (see c.accept_label)
|
|
14
26
|
|
|
15
27
|
jobs:
|
|
16
28
|
duel:
|
|
17
29
|
name: Duel
|
|
18
30
|
runs-on: ubuntu-latest
|
|
31
|
+
# Whether to comment (post, change or none), the run's status (passed,
|
|
32
|
+
# failed or neutral), and whether to mark a push passed (true, or
|
|
33
|
+
# nothing), from Duel.
|
|
34
|
+
outputs:
|
|
35
|
+
comment: ${{ steps.compare.outputs.comment }}
|
|
36
|
+
status: ${{ steps.compare.outputs.status }}
|
|
37
|
+
passed: ${{ steps.compare.outputs.passed }}
|
|
19
38
|
<% if working_directory -%>
|
|
20
39
|
# The app is in a subdirectory of the repository: Duel runs there, and
|
|
21
40
|
# checks out the whole repository for the baseline.
|
|
@@ -27,6 +46,7 @@ jobs:
|
|
|
27
46
|
- uses: actions/checkout@v7
|
|
28
47
|
with:
|
|
29
48
|
fetch-depth: 0 # the baseline comes from history, tags included
|
|
49
|
+
persist-credentials: false
|
|
30
50
|
<% if database == "postgresql" -%>
|
|
31
51
|
# Each worker starts a PostgreSQL server of its own (duel-postgresql).
|
|
32
52
|
- run: sudo apt-get install -y postgresql libpq-dev
|
|
@@ -42,24 +62,59 @@ jobs:
|
|
|
42
62
|
<% end -%>
|
|
43
63
|
<% if version_file? -%>
|
|
44
64
|
# Each version runs on the Ruby its own .ruby-version names (see
|
|
45
|
-
# .duel.rb)
|
|
46
|
-
# too, linked where Duel finds it (as
|
|
47
|
-
# before the app's own Ruby, which runs Duel.
|
|
65
|
+
# .duel.rb). When the baseline's names another Ruby than this code's,
|
|
66
|
+
# that one is installed too, and linked where Duel finds it (as
|
|
67
|
+
# chruby would), before the app's own Ruby, which runs Duel. The
|
|
68
|
+
# baseline is the Compare step's: where HEAD left the branch a pull
|
|
69
|
+
# request targets (or run by hand, the main branch), a merge queue's
|
|
70
|
+
# base, the commit before a push (or the last commit Duel passed, PASSED, if there is
|
|
71
|
+
# one: see the Mark the commit passed job), or on a schedule the last
|
|
72
|
+
# release, the newest tag before HEAD that RELEASE_TAGS matches. (If
|
|
73
|
+
# .duel.rb sets c.main_branch or c.release_tags, set MAIN_BRANCH or
|
|
74
|
+
# RELEASE_TAGS to match.)
|
|
48
75
|
- name: Find the baseline's Ruby
|
|
49
76
|
id: baseline
|
|
50
|
-
|
|
51
|
-
|
|
77
|
+
env:
|
|
78
|
+
BEFORE: ${{ github.event.before }}
|
|
79
|
+
QUEUE_BASE: ${{ github.event.merge_group.base_sha }}
|
|
80
|
+
PASSED: origin/duel/passed<%= "-#{working_directory.tr("/", "-")}" if working_directory %>/${{ github.ref_name }}
|
|
81
|
+
MAIN_BRANCH: origin/${{ github.base_ref || github.event.repository.default_branch }}
|
|
82
|
+
RELEASE_TAGS: "v*"
|
|
83
|
+
run: |
|
|
84
|
+
case "$GITHUB_EVENT_NAME" in
|
|
85
|
+
push)
|
|
86
|
+
base=$BEFORE
|
|
87
|
+
if git merge-base --is-ancestor "$PASSED" HEAD 2>/dev/null; then base=$PASSED; fi ;;
|
|
88
|
+
merge_group) base=$QUEUE_BASE ;;
|
|
89
|
+
schedule) base=$(git describe --tags --abbrev=0 --match "$RELEASE_TAGS" HEAD^ 2>/dev/null) || base= ;;
|
|
90
|
+
*) base=$(git merge-base HEAD "$MAIN_BRANCH" 2>/dev/null) || base= ;;
|
|
91
|
+
esac
|
|
92
|
+
ruby_of() { git show "$1:./.ruby-version" 2>/dev/null | awk 'NF { print $1; exit }'; }
|
|
93
|
+
ruby=
|
|
94
|
+
# (None without a baseline: a branch's first push, or no release yet.)
|
|
95
|
+
if [ -n "$base" ] && git cat-file -e "$base^{commit}" 2>/dev/null; then
|
|
96
|
+
ruby=$(ruby_of "$base")
|
|
97
|
+
[ "$ruby" = "$(ruby_of HEAD)" ] && ruby=
|
|
98
|
+
fi
|
|
99
|
+
echo "ruby=$ruby" >> "$GITHUB_OUTPUT"
|
|
52
100
|
- uses: ruby/setup-ruby@v1
|
|
53
101
|
id: baseline-ruby
|
|
54
102
|
if: steps.baseline.outputs.ruby != ''
|
|
55
103
|
with:
|
|
104
|
+
<% if working_directory -%>
|
|
105
|
+
working-directory: <%= working_directory %>
|
|
106
|
+
<% end -%>
|
|
56
107
|
ruby-version: ${{ steps.baseline.outputs.ruby }}
|
|
108
|
+
# The bundle for this Ruby too, which the baseline's workers load.
|
|
109
|
+
# (Gems installed for one Ruby version don't load on another.)
|
|
110
|
+
bundler-cache: true
|
|
57
111
|
- name: Link the baseline's Ruby where Duel finds it
|
|
58
112
|
if: steps.baseline.outputs.ruby != ''
|
|
113
|
+
env:
|
|
114
|
+
PREFIX: ${{ steps.baseline-ruby.outputs.ruby-prefix }}
|
|
59
115
|
run: |
|
|
60
|
-
prefix="${{ steps.baseline-ruby.outputs.ruby-prefix }}"
|
|
61
116
|
mkdir -p ~/.rubies
|
|
62
|
-
ln -sfn "$
|
|
117
|
+
ln -sfn "$PREFIX" ~/.rubies/ruby-"$("$PREFIX/bin/ruby" -e 'print RUBY_VERSION')"
|
|
63
118
|
<% end -%>
|
|
64
119
|
- uses: ruby/setup-ruby@v1
|
|
65
120
|
with:
|
|
@@ -70,40 +125,122 @@ jobs:
|
|
|
70
125
|
ruby-version: "<%= ruby_version %>"
|
|
71
126
|
<% end -%>
|
|
72
127
|
bundler-cache: true
|
|
73
|
-
#
|
|
74
|
-
#
|
|
75
|
-
#
|
|
76
|
-
|
|
128
|
+
# The labels of the pull request (or for a push, of the one that
|
|
129
|
+
# merged it), for c.accept_label in .duel.rb, as a JSON array. They're
|
|
130
|
+
# read as the job runs, so after adding the label, run it again.
|
|
131
|
+
- name: Find the pull request's labels
|
|
132
|
+
id: labels
|
|
133
|
+
if: github.event_name == 'pull_request' || github.event_name == 'merge_group' || github.event_name == 'push'
|
|
134
|
+
continue-on-error: true # (without them, nothing's accepted)
|
|
135
|
+
env:
|
|
136
|
+
GH_TOKEN: ${{ github.token }}
|
|
137
|
+
PR: ${{ github.event.pull_request.number }}
|
|
138
|
+
QUEUED: ${{ github.event.merge_group.head_ref }}
|
|
139
|
+
run: |
|
|
140
|
+
# (In a merge queue, the last pull request queued, which the queue's
|
|
141
|
+
# branch names: gh-readonly-queue/main/pr-123-<sha>.)
|
|
142
|
+
if [ -n "$QUEUED" ]; then
|
|
143
|
+
PR=${QUEUED##*/pr-}
|
|
144
|
+
PR=${PR%%-*}
|
|
145
|
+
fi
|
|
146
|
+
if [ -n "$PR" ]; then
|
|
147
|
+
labels=$(gh api "repos/$GITHUB_REPOSITORY/issues/$PR/labels" --jq '[.[].name]')
|
|
148
|
+
else
|
|
149
|
+
labels=$(gh api "repos/$GITHUB_REPOSITORY/commits/$GITHUB_SHA/pulls" --jq '[.[0].labels[]?.name]')
|
|
150
|
+
fi
|
|
151
|
+
echo "labels=$labels" >> "$GITHUB_OUTPUT"
|
|
152
|
+
# Against the branch it targets for a pull request, the queue's base
|
|
153
|
+
# for a merge queue, what was there before for a push (or as
|
|
154
|
+
# c.push_base in .duel.rb says, the last commit Duel passed), and the
|
|
155
|
+
# last release on a schedule (slow creep). Duel adds its report to
|
|
156
|
+
# the job's summary, writes it to duel.md for the comment, and sets
|
|
157
|
+
# the step's outputs to say whether to comment, as c.comment in
|
|
158
|
+
# .duel.rb says, whatever the outcome.
|
|
77
159
|
- name: Compare
|
|
160
|
+
id: compare
|
|
78
161
|
env:
|
|
79
|
-
DUEL_BASE: ${{ github.event_name == 'schedule' && 'last-release' || github.event_name == 'push' && github.event.before || '' }}
|
|
162
|
+
DUEL_BASE: ${{ github.event_name == 'schedule' && 'last-release' || github.event_name == 'push' && github.event.before || github.event_name == 'merge_group' && github.event.merge_group.base_sha || '' }}
|
|
80
163
|
# (Run by hand, it compares anyway: with nothing changed, the code
|
|
81
164
|
# with itself, which checks the setup.)
|
|
82
165
|
DUEL_FORCE: ${{ github.event_name == 'workflow_dispatch' && '1' || '' }}
|
|
166
|
+
DUEL_LABELS: ${{ steps.labels.outputs.labels }}
|
|
167
|
+
run: bundle exec duel --markdown "$RUNNER_TEMP/duel/duel.md"
|
|
168
|
+
# For a pull request, what Duel said, for the Perf comment workflow,
|
|
169
|
+
# which can't see this job's outputs.
|
|
170
|
+
- name: Say whether to comment
|
|
171
|
+
if: ${{ !cancelled() && github.event_name == 'pull_request' }}
|
|
172
|
+
env:
|
|
173
|
+
PR: ${{ github.event.pull_request.number }}
|
|
174
|
+
COMMENT: ${{ steps.compare.outputs.comment }}
|
|
175
|
+
STATUS: ${{ steps.compare.outputs.status }}
|
|
83
176
|
run: |
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
177
|
+
mkdir -p "$RUNNER_TEMP/duel"
|
|
178
|
+
printf 'pr=%s\ncomment=%s\nstatus=%s\n' "$PR" "$COMMENT" "$STATUS" > "$RUNNER_TEMP/duel/comment.env"
|
|
179
|
+
# The report, for the comments.
|
|
180
|
+
- name: Keep the report
|
|
181
|
+
if: ${{ !cancelled() && (github.event_name == 'pull_request' || github.event_name == 'push') }}
|
|
182
|
+
uses: actions/upload-artifact@v7
|
|
183
|
+
with:
|
|
184
|
+
name: duel-report
|
|
185
|
+
path: ${{ runner.temp }}/duel/
|
|
186
|
+
if-no-files-found: ignore # none when Duel didn't get to compare
|
|
187
|
+
retention-days: 1
|
|
188
|
+
|
|
189
|
+
# The report, as a new comment on a pushed commit, which notifies its
|
|
190
|
+
# author, when Duel says to, as for a pull request (see the Perf comment
|
|
191
|
+
# workflow). It does only where a regression in a push passes, with
|
|
192
|
+
# c.fail_on = :pull_requests in .duel.rb, so the comment says what the
|
|
193
|
+
# check doesn't. Its last comment is the last one on any commit. This
|
|
194
|
+
# job checks nothing out and runs none of the code.
|
|
195
|
+
commit-comment:
|
|
196
|
+
name: Comment on the commit
|
|
197
|
+
needs: duel
|
|
198
|
+
if: ${{ !cancelled() && github.event_name == 'push' && (needs.duel.outputs.comment == 'post' || needs.duel.outputs.comment == 'change') }}
|
|
199
|
+
runs-on: ubuntu-latest
|
|
200
|
+
permissions:
|
|
201
|
+
contents: write # to comment on the commit with the report
|
|
202
|
+
steps:
|
|
203
|
+
- name: Get the report
|
|
204
|
+
id: report
|
|
205
|
+
continue-on-error: true # there's none when the Duel job failed before comparing
|
|
206
|
+
uses: actions/download-artifact@v8
|
|
207
|
+
with:
|
|
208
|
+
name: duel-report
|
|
209
|
+
path: ${{ runner.temp }}/report
|
|
210
|
+
- name: Comment on the commit
|
|
211
|
+
if: steps.report.outcome == 'success'
|
|
96
212
|
env:
|
|
97
213
|
GH_TOKEN: ${{ github.token }}
|
|
98
|
-
|
|
214
|
+
COMMENT: ${{ needs.duel.outputs.comment }}
|
|
215
|
+
STATUS: ${{ needs.duel.outputs.status }}
|
|
99
216
|
run: |
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
gh api --method PATCH "repos/$GITHUB_REPOSITORY/issues/comments/${ids%%$'\n'*}" -F body=@"$RUNNER_TEMP/comment.md" > /dev/null
|
|
107
|
-
else
|
|
108
|
-
gh api --method POST "repos/$GITHUB_REPOSITORY/issues/$PR/comments" -F body=@"$RUNNER_TEMP/comment.md" > /dev/null
|
|
217
|
+
marker="<!-- duel<%= " #{working_directory}" if working_directory %>"
|
|
218
|
+
last=$(gh api --paginate "repos/$GITHUB_REPOSITORY/comments" \
|
|
219
|
+
--jq ".[] | select(.user.login == \"github-actions[bot]\") | .body | split(\"\n\")[0] | rtrimstr(\"\r\") | select(startswith(\"$marker status=\"))" | tail -n 1)
|
|
220
|
+
if [ "$COMMENT" = post ] || { [ -n "$last" ] && [ "$last" != "$marker status=$STATUS -->" ]; }; then
|
|
221
|
+
{ echo "$marker status=$STATUS -->"; cat "$RUNNER_TEMP/report/duel.md"; } > "$RUNNER_TEMP/comment.md"
|
|
222
|
+
gh api --method POST "repos/$GITHUB_REPOSITORY/commits/$GITHUB_SHA/comments" -F body=@"$RUNNER_TEMP/comment.md" > /dev/null
|
|
109
223
|
fi
|
|
224
|
+
|
|
225
|
+
# With c.push_base = :last_passed in .duel.rb, a push is compared with
|
|
226
|
+
# the last commit Duel passed, which a branch marks (each branch pushed
|
|
227
|
+
# has its own), so slowdowns too small for any one push to fail add up
|
|
228
|
+
# until they fail one. This job moves the branch to a push that passed
|
|
229
|
+
# (or makes it). It checks nothing out and runs none of the code. To
|
|
230
|
+
# accept a regression, move the branch yourself (for main, git push
|
|
231
|
+
# origin main:duel/passed<%= "-#{working_directory.tr("/", "-")}" if working_directory %>/main).
|
|
232
|
+
passed:
|
|
233
|
+
name: Mark the commit passed
|
|
234
|
+
needs: duel
|
|
235
|
+
if: ${{ !cancelled() && github.event_name == 'push' && needs.duel.outputs.passed == 'true' }}
|
|
236
|
+
runs-on: ubuntu-latest
|
|
237
|
+
permissions:
|
|
238
|
+
contents: write # to move the branch
|
|
239
|
+
steps:
|
|
240
|
+
- name: Move the branch to the commit
|
|
241
|
+
env:
|
|
242
|
+
GH_TOKEN: ${{ github.token }}
|
|
243
|
+
BRANCH: duel/passed<%= "-#{working_directory.tr("/", "-")}" if working_directory %>/${{ github.ref_name }}
|
|
244
|
+
run: |
|
|
245
|
+
gh api --method PATCH "repos/$GITHUB_REPOSITORY/git/refs/heads/$BRANCH" -f sha="$GITHUB_SHA" -F force=true > /dev/null 2>&1 ||
|
|
246
|
+
gh api --method POST "repos/$GITHUB_REPOSITORY/git/refs" -f ref="refs/heads/$BRANCH" -f sha="$GITHUB_SHA" > /dev/null
|
metadata
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
|
2
2
|
name: duel-rails
|
|
3
3
|
version: !ruby/object:Gem::Version
|
|
4
|
-
version: 0.1.
|
|
4
|
+
version: 0.1.2
|
|
5
5
|
platform: ruby
|
|
6
6
|
authors:
|
|
7
7
|
- Erik Berlin
|
|
@@ -75,6 +75,7 @@ files:
|
|
|
75
75
|
- lib/generators/duel/install/templates/benchmark.rb.tt
|
|
76
76
|
- lib/generators/duel/install/templates/boot_duel.rb.tt
|
|
77
77
|
- lib/generators/duel/install/templates/duel.rb.tt
|
|
78
|
+
- lib/generators/duel/install/templates/perf-comment.yml.tt
|
|
78
79
|
- lib/generators/duel/install/templates/perf.yml.tt
|
|
79
80
|
- lib/generators/duel/install/templates/requests_duel.rb.tt
|
|
80
81
|
- sig/duel/rails.rbs
|