feat(run): list what a cached run depends on with `vp run -vv` by wan9chi · Pull Request #969 · voidzero-dev/vite-task

wan9chi · 2026-11-07T11:38:06Z

Motivation

When a task hits but shouldn't, the question is whether the file you edited is an input at all. The dry-run RFC's -vv lists everything the cached run depends on, changed or not, and the files a hit restores. This is the last PR of the dry-run stack.

Changes

$ vp run --dry-run -vv vv
…
  [1] app#vv: $ vtt list-dir src
      → Cache hit
        Unchanged: 'src' (listing)               ← new with -vv
  ·······················································
  [3] app#vv: $ vtt stat-file missing.txt
      → Cache hit
        Unchanged: 'missing.txt' (missing)       ← new with -vv
  ·······················································
  [4] app#vv: $ vtt write-file dist/out.txt built
      → Cache hit
        Outputs:   'dist/out.txt'                ← new with -vv

-v becomes a counted flag (ArgAction::Count), so -vv is the same as --verbose --verbose. -vv asks for FileCheck::AllAndBaseline: every file is checked, as with -v, and the baseline is described.
With -vv only, a hit, and the comparison of a miss, carry BaselineDetails, all sorted:
- env names;
- inputs, by path: a file with its hash and whether it may be missing, a folder with its entries when it's listed, or a missing path;
- output files, from archive::list_output_files. It reads the archive without extracting it, with the reader that the remote download's check already used.
Each entry's check builds them, off the scheduler's thread. A dry run doesn't download a remote run's archive, so its outputs read (not downloaded) in the text and "notDownloaded" in the JSON, rather than looking like none. A real run lists them once it has downloaded the archive. A miss after a command, config or env change has neither, since only the previous run's key is compared.
Text: Unchanged: (what didn't change) and Outputs: in the detailed summary. --last-details shows them too.
JSON: baseline.fingerprint, with env and inputs, and baseline.outputs, a list of files or "notDownloaded". The schema change is additive.
Tests: the e2e case remote_cache::very_verbose_remote_hit covers a remote hit with -vv in a dry run, as text and JSON, and in a real run, plus a remote hit without outputs.

🤖 Generated with Claude Code

Note

Low Risk
Additive CLI and reporting behavior gated on -vv; cache lookup logic is extended but existing -v/default paths are unchanged aside from the verbose flag type.

Overview
vp run -vv extends extra-verbose runs so you can see what each cached step actually depends on and what a hit would restore—useful when a task hits but you expected a miss.

-v is now a counted flag: one -v still lists every file change on a miss; -vv adds FileCheck::AllAndBaseline, which builds BaselineDetails (sorted env names, inputs with hashes/listings/missing paths, and output file paths from the cache archive via list_output_files without extracting).

Those details surface in the detailed summary as Unchanged: and Outputs: (including (not downloaded) for remote hits in dry runs), in --last-details / saved summaries, and in --dry-run --json as baseline.fingerprint and baseline.outputs ("notDownloaded" when appropriate). The run-plan JSON schema and docs are updated accordingly; new e2e snapshots cover local and remote -vv behavior.

^{Reviewed by Cursor Bugbot for commit 62c1665. Bugbot is set up for automated code reviews on this repo. Configure here.}

github-actions · 2026-11-07T11:39:56Z

fspy benchmark

linux

dynamic/launch             change  -0.21%  [ -4.76% ..  +5.57%]  overhead   +59.07%
dynamic/access             change  -1.47%  [ -7.71% ..  +4.61%]  overhead   +21.90%
dynamic/access-relative    change  +0.00%  [ -3.45% ..  +4.26%]  overhead   +68.03%
dynamic/access-contended   change  -0.64%  [ -6.61% ..  +7.45%]  overhead   +35.30%
static/launch              change  +0.49%  [ -4.30% ..  +5.83%]  overhead  +196.64%
static/access              change  -0.12%  [ -7.04% ..  +4.68%]  overhead +1443.78%
static/access-relative     change  +0.48%  [ -6.59% ..  +8.54%]  overhead +1982.02%
static/access-contended    change  -0.37%  [ -7.26% ..  +2.58%]  overhead +1362.69%

macos

dynamic/launch             change  +0.25%  [ -2.03% ..  +2.98%]  overhead  +230.95%
dynamic/access             change  +0.00%  [ -3.34% ..  +3.01%]  overhead    +5.40%
dynamic/access-relative    change  -0.22%  [ -4.08% ..  +5.36%]  overhead  +236.29%
dynamic/access-contended   change  -0.44%  [ -6.16% ..  +7.16%]  overhead    +2.37%

windows

dynamic/launch             change  -0.09%  [ -3.13% ..  +3.85%]  overhead   +27.23%
dynamic/access             change  -0.36%  [ -2.14% ..  +1.62%]  overhead    +1.11%
dynamic/access-relative    change  +0.17%  [ -3.13% ..  +2.49%]  overhead    +1.77%
dynamic/access-contended   change  +0.17%  [ -5.13% ..  +2.90%]  overhead    +1.75%

cursor

Stale Bugbot comment from a previous run.

cursor

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 5300b36. Configure here.}

@lifeiscontent

## Motivation Before starting a long run, there's no way to see which tasks would come from the cache and which would run, or why one would miss, without running them (#791, voidzero-dev/vite-plus#2848). [The dry-run RFC](https://github.com/voidzero-dev/vite-task/blob/main/docs/rfcs/dry-run.md) specifies `vp run --dry-run` for this. This is the first PR of the stack that implements it: the text output. The PRs above it add every change of a miss (#961), `-v` (#963), `--json` (#966, #967), `unknown` (#968) and `-vv` (#969). This supersedes #792 by @lifeiscontent, whose write-free lookup and test scenario it builds on, so it credits them as co-author. Closes #791. ## Changes `vp run --dry-run` plans the run and checks the cache as a real run does, the remote cache included, then prints each command's line and the summary. It runs nothing and records nothing. The cache is opened as in a real run, but no entry is written, a hit's outputs aren't downloaded or restored, and `last-summary.json` isn't written, so `--last-details` still shows the last real run. ``` $ vp run --dry-run test $ vtt print-file src/a.txt ◉ cache hit $ vtt print tested --- vp run: 1/2 cache hit (50%), nothing was run. ``` - A dry run goes through the real run's lookup and spawn path, which skips each step that would write or run: recording a local hit, downloading a remote hit, replaying a hit, and spawning the command. - Each line is a real run's, without `, executing` or `, replaying`, and with no blank line between commands. `--log` is ignored, since it only shapes task output: `--dry-run` and `--log` move from `RunFlags`, which selects tasks, to `RunCommand`, and resolve to a `RunMode` of `Real { log }` or `Dry`, so a dry run can't carry a log mode. - The summary follows a real run's rules, with "nothing was run" in place of the successful count and the time saved, and without the `--last-details` hint. The detailed summary (`-v`) has no ✓ or exit codes. Real run, unchanged: ``` vp run: 1/2 cache hit (50%), 2/2 successful. (Run `vp run --last-details` for full details) │ │ │ │ │ │ │ │ │ └─ tasks │ │ │ └─ tasks that succeeded │ │ └─ cache hit rate │ └─ tasks └─ cache hits ``` Dry run: ``` vp run: 1/2 cache hit (50%), nothing was run. │ │ │ │ │ └─ cache hit rate │ └─ tasks └─ cache hits ``` - A cache lookup error is printed under its command and stops the dry run, as it stops a real run, and both exit 1. Real runs stopped but exited 0 before, since the summary only counted tasks that started; a task that fails before it starts now fails the run. Otherwise a dry run exits 0 once planning succeeds. - A nested `vp run --dry-run` isn't inlined. It's started as a process that prints its own dry run, so an enclosing dry run doesn't start it. - Without a task, `--dry-run` is an error, like `vp run -v`. After the task name, it's passed to the task, like any other flag. - RFC: the summary and cache-error rules are written down, and the sentence about the first changed file being the first in path order is removed. Without `-v`, the file check stops at whichever changed file it finds first. 🤖 Generated with [Claude Code](https://claude.com/claude-code)  --- > [!NOTE] > **Medium Risk** > Touches cache lookup, remote fetch, spawn, and exit-code paths across the runner; mistakes could skip real execution side effects or mis-report failures, though behavior is gated on `RunKind::Dry`. > > **Overview** > Adds **`vp run --dry-run`**, which plans the task graph and performs full cache lookups (including remote) but **does not** spawn processes, replay outputs, download remote archives, write cache entries, or persist `last-summary.json`. > > The CLI introduces **`RunMode`/`RunKind`**: `--dry-run` and `--log` move onto `RunCommand` (dry runs always use interleaved output). Execution and cache layers thread `RunKind` through lookup, spawn, and reporters so dry runs short-circuit after reporting cache status (`WouldSpawn` / `DryRun` not-updated reasons). > > **Output and summary** match a real run’s command lines and cache indicators, but omit `, executing` / `, replaying`, skip extra blank lines, use **“nothing was run”** instead of success/saved time, and omit the `--last-details` hint. **`-v`** detailed summaries drop exit icons and saved-time wording for hits. > > **Bugfix:** runs now **exit 1** when a task’s cache cannot be checked (e.g. DB unreadable), including failures before `start()`; nested `vp run --dry-run` is not inlined. Changelog and dry-run RFC/e2e snapshots document the behavior. > > Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit 3010605. Bugbot is set up for automated code reviews on this repo. Configure [here](https://www.cursor.com/dashboard/bugbot).  Co-authored-by: Aaron Reisman <aaron@lifeiscontent.net> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

When a task hits but shouldn't, the question is whether the edited file is an input at all. With `-vv`, the detailed summary and the JSON list everything the cached run depends on, changed or not, and the files it restores. - `-v` becomes counted (`ArgAction::Count`), and `-vv` asks for `FileCheck::AllAndBaseline`: every file is checked, as with `-v`, and the baseline is described. - With `-vv` only, a hit and a miss's comparison carry `BaselineDetails`: env names, inputs by path (a file with its hash and whether it may be missing, a folder with its entries when listed, or a missing path), and output files from `archive::list_output_files`, which reads the archive without extracting it, with the reader the remote download's check already used. All of it is sorted. Each entry's check builds it, off the scheduler's thread. A dry run doesn't download a remote run's archive, so its outputs are `BaselineOutputs::NotDownloaded`: `(not downloaded)` in the text and `"notDownloaded"` in the JSON. A real run lists them once it has downloaded the archive. - Text: `Unchanged:` (what didn't change) and `Outputs:` in the detailed summary, which `--last-details` shows too. - JSON: `baseline.fingerprint`, with `env` and `inputs`, and `baseline.outputs`. The schema change is additive. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…proxy (#985) ## Motivation `remote_cache::upload_without_token` failed on #969's macOS CI ([job](https://github.com/voidzero-dev/vite-task/actions/runs/42608444639/job/129130427409)). The test expects the backend's `401 Invalid credentials` for an upload without a token. Instead `vp run` got `HTTP status 502`, and `remote-cache-server stop` reported `POST /store failed: write EPIPE`. The test predates the dry-run stack; the race is in the e2e proxy that sits in front of the backend. ## Changes - The proxy streamed each request body to the backend as it arrived (`request.pipe(upstream)`). The Worker rejects an upload by its headers and answers 401 before reading the body. workerd then closes the connection, resetting it when body bytes are still unread. - On a slow runner, a later piece of the body reached the proxy after that reset but before the proxy had read the 401. Writing it failed with EPIPE, Node destroyed the socket with the 401 still unread, and the proxy answered 502. - The proxy now reads the whole request first and sends it with `upstream.end(body)`. The body then goes out together with the headers, before the backend can answer, and an error after the answer is ignored, as before. 🤖 Generated with [Claude Code](https://claude.com/claude-code)  --- > [!NOTE] > **Low Risk** > Changes only the e2e remote-cache proxy’s request forwarding; production cache paths are unaffected. > > **Overview** > Fixes a race in the e2e remote-cache **proxy** where uploads without a token were reported as **502** instead of the backend’s early **401**. > > The tap no longer streams the client body with `request.pipe(upstream)`. It **buffers** the full request, then sends it with `upstream.end(Buffer.concat(body))` so headers and body reach the Worker before it can reject and reset the connection. Comments now describe that reset-after-answer behavior and why late body writes could drop the unread 401. > > Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit e63c3be. Bugbot is set up for automated code reviews on this repo. Configure [here](https://www.cursor.com/dashboard/bugbot).  Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

…it (#986) ## Motivation `dry_run_json::unknown_after_remote_hit`, added in #968, is flaky. It failed on main's macOS CI ([job](https://github.com/voidzero-dev/vite-task/actions/runs/42609456309/job/129134222870)), on #969's, and on #984's Windows job ([run](https://github.com/voidzero-dev/vite-task/actions/runs/42609782121)). In its first step, the backend log read `fetch 404, store 200, fetch 404, store 200` instead of `fetch 404, fetch 404, store 200, store 200`. ## Changes - **Cause:** `vt run after-emit` runs `emit` first. Once `emit` finishes, its upload starts in the background at the same time as `after-emit`'s fetch, and the e2e proxy logs each request when its response finishes, so either can come first. `VP_RUN_INTERNAL_HIDE_PENDING_UPLOADS` hides the upload's progress, but doesn't delay it. - **Fix:** each dependency now runs in its own step first: `vt run emit` before `vt run after-emit`, and `vt run build` before `vt run after-build`. Each step then makes at most one fetch followed by one upload, in a fixed order. - The dry runs that the case checks are unchanged: `after-emit` is still `unknown` after a remote hit that restores outputs, and `after-build` is still a hit. 🤖 Generated with [Claude Code](https://claude.com/claude-code)  --- > [!NOTE] > **Low Risk** > Test fixture and snapshot updates only; no production run or remote-cache logic changes. > > **Overview** > Fixes flaky **`unknown_after_remote_hit`** dry-run e2e snapshots by **splitting combined runs into separate steps**: `vt run emit` before `vt run after-emit`, and `vt run build` before `vt run after-build`. > > That keeps each remote-cache interaction to a **deterministic fetch-then-store order** (uploads from the first task no longer overlap the dependent task’s fetch when the proxy logs responses). Comments and the **`unknown_after_remote_hit.md`** snapshot were updated for the extra steps, cache-hit replay lines, and corrected expectations (e.g. `after-build` as full cache hit with **no fetch**). > > **`snapshots.toml`** and the markdown snapshot are the only touched files; asserted dry-run behavior is unchanged. > > Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit 9226b86. Bugbot is set up for automated code reviews on this repo. Configure [here](https://www.cursor.com/dashboard/bugbot).  Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

wan9chi added this pull request to stack #970 November 7, 2026 11:38

wan9chi mentioned this pull request Nov 7, 2026

feat(run): preview a run with vp run --dry-run #960

Merged

cursor Bot reviewed Nov 7, 2026

View reviewed changes

Comment thread crates/vt/src/session/reporter/summary.rs Resolved

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from a868c8f to 5300b36 Compare November 7, 2026 11:50

cursor Bot reviewed Nov 7, 2026

View reviewed changes

Comment thread crates/vt/src/session/reporter/summary.rs Resolved

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from 5300b36 to 9addbeb Compare November 7, 2026 12:00

wan9chi removed this pull request from stack #970 November 7, 2026 14:54

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from 9addbeb to 2169ef1 Compare November 7, 2026 14:54

wan9chi added this pull request to stack #973 November 7, 2026 14:55

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from 2169ef1 to ac3c017 Compare November 7, 2026 15:07

wan9chi removed this pull request from stack #973 November 7, 2026 17:18

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from ac3c017 to 095b9ef Compare November 7, 2026 17:19

wan9chi added this pull request to stack #975 November 7, 2026 17:19

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from 095b9ef to e9fc051 Compare November 7, 2026 18:04

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from e9fc051 to b1bd8ed Compare November 7, 2026 18:38

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from b1bd8ed to 8d3d5c4 Compare November 9, 2026 10:11

wan9chi force-pushed the claude/dry-run-11-very-verbose branch 2 times, most recently from 20447dd to 38391a9 Compare November 9, 2026 11:21

Base automatically changed from claude/dry-run-10-unknown to main November 9, 2026 11:29

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from 38391a9 to 7cd8aaf Compare November 9, 2026 11:29

wan9chi removed this pull request from stack #975 November 9, 2026 12:03

This was referenced Nov 9, 2026

test(cache): keep the backend's early answer to an upload in the e2e proxy #985

Merged

test(run): fix the order of remote requests in unknown_after_remote_hit #986

Merged

wan9chi force-pushed the claude/dry-run-11-very-verbose branch from 7cd8aaf to 62c1665 Compare November 9, 2026 12:04

wan9chi changed the base branch from main to claude/e2e-remote-unknown-order November 9, 2026 12:04

wan9chi added this pull request to stack #987 November 9, 2026 12:04

wan9chi merged commit dc54330 into main Nov 9, 2026
22 checks passed

wan9chi deleted the claude/dry-run-11-very-verbose branch November 9, 2026 12:56

Navigation Menu

Repository navigation

feat(run): list what a cached run depends on with `vp run -vv` - #969

feat(run): list what a cached run depends on with `vp run -vv`#969
wan9chi merged 1 commit into
claude/e2e-remote-unknown-orderfrom
claude/dry-run-11-very-verbose

wan9chi commented Nov 7, 2026 •

edited by cursor Bot

Loading

Uh oh!

github-actions Bot commented Nov 7, 2026 •

edited

Loading

Uh oh!

cursor Bot left a comment •

edited

Loading

Uh oh!

Uh oh!

cursor Bot left a comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

wan9chi commented Nov 7, 2026 • edited by cursor Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Motivation

Changes

Uh oh!

github-actions Bot commented Nov 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

fspy benchmark

linux

macos

windows

Uh oh!

cursor Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

wan9chi commented Nov 7, 2026 •

edited by cursor Bot

Loading

github-actions Bot commented Nov 7, 2026 •

edited

Loading

cursor Bot left a comment •

edited

Loading