Skip to content

Commit ec2750a

Browse files
committed
Revert "Measure mounted REPL typing latency on affected pull requests"
This reverts commit 2745925.
1 parent 6cc1a3e commit ec2750a

7 files changed

Lines changed: 1 addition & 418 deletions

File tree

‎.github/workflows/ci.yml‎

Lines changed: 0 additions & 38 deletions
Original file line numberDiff line numberDiff line change
@@ -198,44 +198,6 @@ jobs:
198198
repl bench checkout/.bench/*/closed.json
199199
include-hidden-files: "true"
200200

201-
- name: Compare mounted REPL typing on the PR revisions
202-
if: matrix.host == 'ubuntu-latest' && github.event_name == 'pull_request'
203-
id: typing
204-
working-directory: repl bench checkout
205-
env:
206-
BASE_REVISION: ${{ github.event.pull_request.base.sha }}
207-
HEAD_REVISION: ${{ github.event.pull_request.head.sha }}
208-
run: |
209-
set -euo pipefail
210-
git fetch --no-tags --depth=1 origin "$BASE_REVISION" "$HEAD_REVISION"
211-
changed=$(git diff --no-renames --name-only "$BASE_REVISION" "$HEAD_REVISION" -- \
212-
packages/ examples/repl-bench/ deno.json deno.lock package.json \
213-
pnpm-lock.yaml bun.lock scripts/setup.ts scripts/deps.ts scripts/lib/ \
214-
.github/workflows/ci.yml)
215-
if [[ -z "$changed" ]]; then
216-
exit 0
217-
fi
218-
printf '%s\n' 'affected=true' >> "$GITHUB_OUTPUT"
219-
scratch=$(mktemp -d "$RUNNER_TEMP/repl-typing-worktrees.XXXXXXXX")
220-
trap 'git worktree remove --force "$scratch/candidate" || true; git worktree remove --force "$scratch/baseline" || true; rmdir "$scratch"' EXIT
221-
git worktree add --detach "$scratch/baseline" "$BASE_REVISION"
222-
git worktree add --detach "$scratch/candidate" "$HEAD_REVISION"
223-
(cd "$scratch/baseline" && deno task setup)
224-
(cd "$scratch/candidate" && deno task setup)
225-
report="$RUNNER_TEMP/repl-typing-latency/${GITHUB_RUN_ATTEMPT}.json"
226-
deno run --allow-all --cached-only --frozen --node-modules-dir=none \
227-
examples/repl-bench/typing-latency.ts "$scratch/baseline" "$scratch/candidate" "$report" \
228-
| tee -a "$GITHUB_STEP_SUMMARY"
229-
(cd "$scratch/candidate" && deno task test packages/cli/tests/repl-sidekick-typing.test.ts)
230-
231-
- uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7
232-
if: ${{ always() && steps.typing.outputs.affected == 'true' }}
233-
with:
234-
name: repl-typing-latency-attempt-${{ github.run_attempt }}
235-
path: ${{ runner.temp }}/repl-typing-latency/
236-
if-no-files-found: error
237-
retention-days: "7"
238-
239201
# The same command the release publishes with, minus --dry-run. Catches slow
240202
# types and unresolvable specifiers before they reach a tag.
241203
jsr:

‎examples/repl-bench/README.md‎

Lines changed: 1 addition & 22 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@ Use the ReplBench plugin to verify what a person sees in the REPL. Start from
44
a tooling checkout prepared with `deno task setup`. The bench uses Bun with
55
PTY support to drive the REPL.
66

7-
[Visual comparison](#compare-repl-builds-with-screenshots) · [UI handoff](#ui-handoff) · [Typing latency](#compare-mounted-typing-latency) · [Tests](#run-the-tests) · [Plugin components](#plugin-components)
7+
[Visual comparison](#compare-repl-builds-with-screenshots) · [UI handoff](#ui-handoff) · [Tests](#run-the-tests) · [Plugin components](#plugin-components)
88

99
## Compare REPL builds with screenshots
1010

@@ -185,27 +185,6 @@ dirty status, working directory, and dimensions. Failed launches keep their
185185
available evidence. `Terminal.Screenshot` saves the visible viewport and
186186
binds its path. Use separate names for the checkpoints you compare.
187187

188-
## Compare mounted typing latency
189-
190-
Prepare a clean baseline and candidate worktree with `deno task setup`. Run
191-
this command from the tooling checkout and give it a path for the JSON report.
192-
193-
```text
194-
deno run --allow-all --cached-only --frozen --node-modules-dir=none \
195-
examples/repl-bench/typing-latency.ts /path/to/baseline /path/to/candidate /tmp/typing.json
196-
```
197-
198-
Both measurements use the mounted REPL at 200×52. Each starts when the test
199-
dispatches one character and ends when a completed emulator write shows the
200-
expected new Message value. The small and 52-round wrapping histories are
201-
generated with a scripted provider before sampling. The report retains all
202-
40 edits per workload and calculates median and p95 after the first 10.
203-
204-
The Linux bench job attaches the paired report for pull requests that change
205-
REPL code, dependencies or tooling. It also runs the deterministic typing and
206-
render-work regression. Compare both revisions in each workload when
207-
investigating a timing change. Shared runners report deltas for review.
208-
209188
## Run the tests
210189

211190
```sh

‎examples/repl-bench/lib/typing-latency-report.ts‎

Lines changed: 0 additions & 114 deletions
This file was deleted.

‎examples/repl-bench/tests/mounted-typing.test.ts‎

Lines changed: 0 additions & 90 deletions
This file was deleted.

‎examples/repl-bench/tests/typing-latency-report.test.ts‎

Lines changed: 0 additions & 50 deletions
This file was deleted.

0 commit comments

Comments
 (0)