@redue/cli 0.1.0-alpha.25 → 0.1.0-alpha.27
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.
- package/CHANGELOG.md +84 -0
- package/README.md +27 -12
- package/bin/redue.mjs +77 -10
- package/docs/README.md +1 -0
- package/docs/user/node-script.md +10 -6
- package/docs/user/sandbox.md +221 -0
- package/docs/user/supported-projects.md +9 -5
- package/package.json +1 -1
- package/src/claude-hook.mjs +60 -14
- package/src/cli.mjs +136 -16
- package/src/daemon.mjs +51 -6
- package/src/decision-validation.mjs +2 -1
- package/src/declared-plan.mjs +10 -1
- package/src/index.mjs +37 -3
- package/src/landlock.py +228 -0
- package/src/node-onboarding.mjs +41 -20
- package/src/node-script.mjs +16 -1
- package/src/presentation.mjs +2 -1
- package/src/python-onboarding.mjs +87 -0
- package/src/sandbox.mjs +493 -0
- package/src/setup.mjs +7 -2
- package/src/stats.mjs +72 -0
package/CHANGELOG.md
CHANGED
|
@@ -1,5 +1,89 @@
|
|
|
1
1
|
# Changelog
|
|
2
2
|
|
|
3
|
+
## 0.1.0-alpha.27 — 2026-11-01
|
|
4
|
+
|
|
5
|
+
A build or test run no longer makes a sandboxed typecheck or lint STALE,
|
|
6
|
+
and `redue stats` shows what REDUE saved you.
|
|
7
|
+
|
|
8
|
+
- **Files git ignores are hidden from sandboxed runs.** A sandboxed
|
|
9
|
+
typecheck or lint now runs without the files git ignores (`dist/`,
|
|
10
|
+
`.next/`, coverage files, `.env`), except `node_modules`, `.venv` and
|
|
11
|
+
runner caches. Those files are not its inputs, so a build or test run that
|
|
12
|
+
writes them leaves the result CURRENT. The claim: the result holds for a
|
|
13
|
+
run that sees only the project's files that git doesn't ignore, its
|
|
14
|
+
installed packages and the toolchain. Only the project's `.gitignore`
|
|
15
|
+
files and `.git/info/exclude` count, not your global excludes file.
|
|
16
|
+
- **Checks that need ignored files keep working.** `init` tries the
|
|
17
|
+
sandbox with ignored files hidden first (`sandbox-v2`), then with the
|
|
18
|
+
whole checkout (`sandbox-v1`), then the general recipe. A hidden run that
|
|
19
|
+
fails on a hidden file runs again in the sandbox without hiding, and that
|
|
20
|
+
result is kept. Checks set up by alpha.26 stay `sandbox-v1` until you run
|
|
21
|
+
`redue init` again.
|
|
22
|
+
- **Measured.** In the product census, after each project's own build and
|
|
23
|
+
tests ran, checks that stayed CURRENT (alpha.26 in brackets, from the
|
|
24
|
+
measurement census): TypeScript on Linux 31 of 37 (5 of 43), on macOS 24
|
|
25
|
+
of 34 (5 of 38); Python on Linux 31 of 38 (24 of 37), on macOS 28 of 34
|
|
26
|
+
(20 of 33). Where a check qualifies with both, the sandbox stayed CURRENT
|
|
27
|
+
16 times and the general recipe 9 of 23 (Linux). The strace cross-check
|
|
28
|
+
found no read outside the recorded inputs.
|
|
29
|
+
- **`redue stats`.** Per check: reruns the hook skipped, the time that
|
|
30
|
+
saved, recorded runs, and the top reasons a check was UNVERIFIED, from a
|
|
31
|
+
log in REDUE's local state.
|
|
32
|
+
- `redue remove-state` no longer fails on Windows while the observer is
|
|
33
|
+
still closing its files.
|
|
34
|
+
- Known limits: a sandboxed run can still see that a hidden file exists
|
|
35
|
+
(its name on macOS; listings, sizes and times on Linux), so a tool that
|
|
36
|
+
acts on existence alone is not covered; tests do not use hidden files for
|
|
37
|
+
that reason. On Linux, a file a run creates in a folder that holds hidden
|
|
38
|
+
files cannot be read back in the same run.
|
|
39
|
+
- Contributors: eval entry points that run third-party code refuse to start
|
|
40
|
+
unless `GITHUB_ACTIONS=true`, and importing any eval entry point throws
|
|
41
|
+
before it does anything.
|
|
42
|
+
|
|
43
|
+
## 0.1.0-alpha.26 — 2026-10-31
|
|
44
|
+
|
|
45
|
+
Typecheck and lint checks run inside an OS sandbox, so REDUE can reuse them
|
|
46
|
+
for native tools and for Python. Python projects are supported for the first
|
|
47
|
+
time.
|
|
48
|
+
|
|
49
|
+
- **Sandboxed recipe.** On macOS (Seatbelt) and Linux 6.7+ (Landlock), a
|
|
50
|
+
root typecheck or lint script runs in a sandbox that sees only the
|
|
51
|
+
project, its installed packages and the toolchain, the way CI does. Its
|
|
52
|
+
inputs are the whole checkout, git-ignored files included, plus the
|
|
53
|
+
installed packages and a fingerprint of the environment it was given. A
|
|
54
|
+
result holds for a run that sees only those. `init` tries the precise
|
|
55
|
+
TypeScript recipes first, then the sandbox, then the general recipe. See
|
|
56
|
+
[sandboxed recipe](docs/user/sandbox.md).
|
|
57
|
+
- **Python.** `redue setup claude` and `init` find mypy, pyright, ty and
|
|
58
|
+
ruff in the project's `.venv`, from its CI workflows, Makefile, justfile,
|
|
59
|
+
tox, nox and scripts. The Claude Code hook recognizes `uv run`,
|
|
60
|
+
`poetry run`, `python -m` and `.venv/bin` forms.
|
|
61
|
+
- **What the sandbox changes.** Network is off. HOME and the temp folder
|
|
62
|
+
are private to the check; HOME is kept between runs so caches persist.
|
|
63
|
+
Tools can read `.git`, but git commands find no repository. A background
|
|
64
|
+
process the check leaves running is stopped when the run ends.
|
|
65
|
+
- **Checked at init.** `init` runs each check plainly and sandboxed, and
|
|
66
|
+
proposes the sandbox only when exit status and output (timings and paths
|
|
67
|
+
aside) match. Otherwise the general recipe applies. A failing run whose
|
|
68
|
+
output looks like a sandbox refusal is rerun plainly; if the plain run
|
|
69
|
+
exits differently, the result is UNVERIFIED. Other failures run once.
|
|
70
|
+
- **Coverage.** In a census of 120 public TypeScript repositories (86
|
|
71
|
+
installed on macOS), 35 have a qualifying check (alpha.24 census: 30):
|
|
72
|
+
typecheck 22, lint 25, tests 2. Of 120 Python repositories, 23 of the 75
|
|
73
|
+
that installed on Linux have one (typecheck 15, lint 22); 20 of 64 on
|
|
74
|
+
macOS. On Linux, every sandboxed run was traced with strace: each file
|
|
75
|
+
it read in the checkout was a recorded input, a cache or its own output.
|
|
76
|
+
- **Measured with Claude Code.** On starlette, graphon and
|
|
77
|
+
claude-agent-sdk-python, sessions took 58 s with REDUE and 125 s without.
|
|
78
|
+
REDUE skipped 10 typechecks and every one matched a plain rerun.
|
|
79
|
+
- Known limits: inside Claude Code's own macOS sandbox and on Windows, the
|
|
80
|
+
sandbox can't start and the general recipe applies. On Linux before 6.7,
|
|
81
|
+
sandboxed runs are UNVERIFIED.
|
|
82
|
+
A build or test run that writes into the checkout (`dist/`, `.next/`,
|
|
83
|
+
coverage files) makes a sandboxed check STALE, as it does a
|
|
84
|
+
general-recipe check: about 1 in 4 TypeScript and 2 in 3 Python checks
|
|
85
|
+
stayed CURRENT through the project's build and tests.
|
|
86
|
+
|
|
3
87
|
## 0.1.0-alpha.25 — 2026-10-29
|
|
4
88
|
|
|
5
89
|
Running your tests no longer makes a typecheck STALE, and `status --sync`
|
package/README.md
CHANGED
|
@@ -10,9 +10,11 @@ check that still holds is not run again and one that is stale is.
|
|
|
10
10
|
|
|
11
11
|
## Quickstart (about 5 minutes)
|
|
12
12
|
|
|
13
|
-
You need macOS or Linux, Node 22 or 24, Git, Claude Code, and a
|
|
14
|
-
TypeScript project with its dependencies installed
|
|
15
|
-
`pnpm install` or `yarn`)
|
|
13
|
+
You need macOS or Linux, Node 22 or 24, Git, Claude Code, and either a
|
|
14
|
+
JavaScript or TypeScript project with its dependencies installed
|
|
15
|
+
(`npm install`, `pnpm install` or `yarn`), or a Python project with a `.venv`
|
|
16
|
+
in its root (`uv sync`, or Poetry with in-project virtualenvs). On Linux, also
|
|
17
|
+
Python 3.
|
|
16
18
|
|
|
17
19
|
```sh
|
|
18
20
|
npm install --global @redue/cli@alpha
|
|
@@ -57,26 +59,39 @@ what the agent asks. Plain `redue status` is a fast cached answer and can say
|
|
|
57
59
|
UNVERIFIED ("cannot confirm your execution context") where `--sync` would say
|
|
58
60
|
CURRENT.
|
|
59
61
|
|
|
62
|
+
`redue stats` shows, per check, the reruns the hook skipped, the time saved,
|
|
63
|
+
the recorded runs and the top reasons a check was UNVERIFIED, from a log kept
|
|
64
|
+
on your machine.
|
|
65
|
+
|
|
60
66
|
To undo everything setup did: `redue remove`.
|
|
61
67
|
|
|
62
68
|
## What it covers today
|
|
63
69
|
|
|
64
|
-
In a census of 120 active public TypeScript repositories (
|
|
65
|
-
at commit
|
|
66
|
-
with dependencies' install scripts on,
|
|
67
|
-
when two recorded runs on the same code both came out CURRENT
|
|
68
|
-
inputs.
|
|
70
|
+
In a census of 120 active public TypeScript repositories (November 2026, REDUE
|
|
71
|
+
at commit e264a67, before 0.1.0-alpha.27), each installed as its users would,
|
|
72
|
+
with dependencies' install scripts on, 90 installed cleanly on a macOS runner.
|
|
73
|
+
A check counts when two recorded runs on the same code both came out CURRENT
|
|
74
|
+
on the same inputs. 35 repositories had at least one such check:
|
|
69
75
|
|
|
70
76
|
| Check | Repositories | How |
|
|
71
77
|
|---|---|---|
|
|
72
|
-
| Typecheck |
|
|
73
|
-
| Lint |
|
|
78
|
+
| Typecheck | 22 | 16 by running the project's script in an OS sandbox ([sandboxed recipe](docs/user/sandbox.md)), 15 of them with the files git ignores hidden; 5 by running the local `tsc` directly ([direct recipe](docs/user/direct-typescript.md)); 1 by running the script with every Node process recorded ([general recipe](docs/user/node-script.md)) |
|
|
79
|
+
| Lint | 24 | 21 sandboxed recipe (19 with ignored files hidden), 3 general recipe |
|
|
74
80
|
| Tests | 2 | Vitest run directly ([direct test recipe](docs/user/tests-direct.md)) |
|
|
75
81
|
|
|
82
|
+
In a census of 120 Python repositories, 74 installed on a Linux runner
|
|
83
|
+
and 23 had a qualifying check: typecheck (mypy, pyright, ty) 15, lint (ruff)
|
|
84
|
+
22, all through the sandboxed recipe with ignored files hidden. A check whose
|
|
85
|
+
git-ignored files are hidden stays CURRENT when a build or test run rewrites
|
|
86
|
+
them: after each project's build and tests ran, 31 of 37 such TypeScript
|
|
87
|
+
checks and 31 of 38 Python checks were still CURRENT on Linux. On Linux, every sandboxed run was traced
|
|
88
|
+
with strace: each file it read in the checkout was a recorded input, a cache
|
|
89
|
+
or its own output.
|
|
90
|
+
|
|
76
91
|
Everything else stays UNVERIFIED, with a reason you can read in
|
|
77
92
|
`redue explain`. The usual ones: a script starts a program REDUE cannot see
|
|
78
|
-
into
|
|
79
|
-
the project uses Bun, or a
|
|
93
|
+
into and the sandbox can't run (inside Claude Code's macOS sandbox, on
|
|
94
|
+
Windows), the project uses Bun, or a check writes into its own inputs.
|
|
80
95
|
[Supported projects](docs/user/supported-projects.md) has the details.
|
|
81
96
|
|
|
82
97
|
## Measured with Claude Code
|
package/bin/redue.mjs
CHANGED
|
@@ -7,6 +7,7 @@ import {directCompiler,directQualification,directEnvironment,directInstalledInpu
|
|
|
7
7
|
import {runArgv,tscRun} from '../src/typescript-invocation.mjs';
|
|
8
8
|
import {testEnvironment,testRunnerAt,testsInterpretation,testsQualification} from '../src/test-direct.mjs';
|
|
9
9
|
import {scriptEnvironment,scriptInterpretation,scriptKinds,scriptQualification} from '../src/node-script.mjs';
|
|
10
|
+
import {hiddenListIssue,hiddenQualification,isSandboxed,pythonTool,sandboxEnvironment,sandboxInterpretationOf,sandboxKinds,sandboxQualification} from '../src/sandbox.mjs';
|
|
10
11
|
import os from 'node:os';
|
|
11
12
|
import path from 'node:path';
|
|
12
13
|
import {createHash} from 'node:crypto';
|
|
@@ -31,7 +32,7 @@ Commands:
|
|
|
31
32
|
setup claude [--yes] Set up REDUE and Claude Code in this project in one step:
|
|
32
33
|
init, agent setup claude, start, one recorded run per check
|
|
33
34
|
remove [--yes] Undo setup: integration, observer state, and the config setup created
|
|
34
|
-
init [--dry-run] [--workspace NAME] [--check KIND] [--recipe typescript-direct|tests-direct|node-script]
|
|
35
|
+
init [--dry-run] [--workspace NAME] [--check KIND] [--recipe typescript-direct|tests-direct|node-script|sandbox]
|
|
35
36
|
Discover existing scripts; optionally select one workspace/check
|
|
36
37
|
agent setup codex|claude [--dry-run|--apply] [--json]
|
|
37
38
|
agent doctor [codex|claude] [--json]
|
|
@@ -45,6 +46,7 @@ Commands:
|
|
|
45
46
|
detail [CHECK] [--json] Alias for explain
|
|
46
47
|
run CHECK [--if-due] Execute one configured check and record its outcome; with --if-due,
|
|
47
48
|
only when its last pass does not apply to the files as they are now
|
|
49
|
+
stats [--json] Per check: skips, time saved, recorded runs, why checks were UNVERIFIED
|
|
48
50
|
remove-state Stop and remove only this configuration's owned local state
|
|
49
51
|
|
|
50
52
|
No checks run in the background. Direct commands outside redue create no receipts.`;
|
|
@@ -94,7 +96,7 @@ if(command==='agent'){
|
|
|
94
96
|
catch(e){error(e.message);}
|
|
95
97
|
await exit(0);
|
|
96
98
|
}
|
|
97
|
-
if(!['init','start','stop','status','detail','explain','run','remove-state'].includes(command))
|
|
99
|
+
if(!['init','start','stop','status','detail','explain','run','stats','remove-state'].includes(command))
|
|
98
100
|
error(`unknown command ${command||'<none>'}; use --help`);
|
|
99
101
|
const json=args.includes('--json');
|
|
100
102
|
// run --if-due: run the check only when its last pass does not apply now.
|
|
@@ -119,9 +121,28 @@ if(short&&(command!=='status'||json||sync))error('--short is a cached status opt
|
|
|
119
121
|
if(dryRun&&command!=='init')error('--dry-run is for init');
|
|
120
122
|
|
|
121
123
|
if(command==='init'){
|
|
122
|
-
const {discoverNodeProject}=await import('../src/node-onboarding.mjs');
|
|
123
|
-
|
|
124
|
-
|
|
124
|
+
const {discoverNodeProject,sourceLinkIssue}=await import('../src/node-onboarding.mjs');
|
|
125
|
+
const {discoverPythonProject}=await import('../src/python-onboarding.mjs');
|
|
126
|
+
const {sandboxTrial}=await import('../src/sandbox.mjs');
|
|
127
|
+
// A Node project, a Python project, or both in one checkout.
|
|
128
|
+
const python=!recipeSelection||recipeSelection==='sandbox'?discoverPythonProject(process.cwd()):null;
|
|
129
|
+
let discovery;
|
|
130
|
+
if(fs.existsSync('package.json')||!python)try{discovery=discoverNodeProject(process.cwd(),{recipe:recipeSelection});}
|
|
131
|
+
catch(e){error(e.message);}
|
|
132
|
+
else{const root=realObservedPath(process.cwd());
|
|
133
|
+
const git=spawnSync('git',['rev-parse','--show-toplevel'],{cwd:root,encoding:'utf8',timeout:3000,stdio:['ignore','pipe','ignore']});
|
|
134
|
+
discovery={schema:1,root,manager:{name:null,version:null,layout:null},observationIssue:git.status===0?sourceLinkIssue(root):
|
|
135
|
+
'Git checkout is unavailable; filesystem observation needs a checkout',workspaces:false,node:process.version,
|
|
136
|
+
checks:[],ambiguous:[],skipped:[],config:{schema:1,checks:[]}};}
|
|
137
|
+
if(python){
|
|
138
|
+
// A Python check takes its tool's name where the name is taken.
|
|
139
|
+
const taken=new Set(discovery.checks.map(row=>row.config.name));
|
|
140
|
+
for(const row of python.checks){if(taken.has(row.config.name))row.config.name=`${row.config.kind}:${row.config.command[0].split('/').at(-1)}`;
|
|
141
|
+
taken.add(row.config.name);}
|
|
142
|
+
if(!discovery.observationIssue)discovery.checks.push(...python.checks);
|
|
143
|
+
discovery.skipped.push(...python.skipped);
|
|
144
|
+
discovery.python=true;
|
|
145
|
+
}
|
|
125
146
|
if(workspaceSelection||checkSelection){
|
|
126
147
|
const chosen=discovery.checks.filter(row=>
|
|
127
148
|
(!workspaceSelection||row.config.workspace===workspaceSelection||
|
|
@@ -130,6 +151,35 @@ if(command==='init'){
|
|
|
130
151
|
if(!chosen.length)error('no discovered check matches the selected workspace and kind');
|
|
131
152
|
discovery.checks=chosen;discovery.config.checks=chosen.map(row=>row.config);
|
|
132
153
|
}
|
|
154
|
+
// A sandboxed check is proposed only when it gives the same exit status and
|
|
155
|
+
// output inside the sandbox as outside it: first with the checkout's
|
|
156
|
+
// git-ignored files hidden (sandbox-v2), else with the whole checkout
|
|
157
|
+
// (sandbox-v1), reusing the plain run. Otherwise init writes what it would
|
|
158
|
+
// have without the sandboxed recipe, with the reason.
|
|
159
|
+
const trialIssue=trial=>trial.issue?`the sandbox cannot run here (${trial.issue})`:
|
|
160
|
+
// A failure both ways that shows a refusal cannot be told from a real one.
|
|
161
|
+
trial.same?null:trial.plain!==trial.sandboxed?`exit ${trial.sandboxed} inside the sandbox, ${trial.plain} outside it${trial.line?` (${trial.line})`:''}`:
|
|
162
|
+
trial.line?`it fails both ways (exit ${trial.plain}), with a sign of a refusal inside the sandbox (${trial.line})`:
|
|
163
|
+
`the output differs inside the sandbox: ${JSON.stringify(trial.output.sandboxed)} instead of ${JSON.stringify(trial.output.plain)}`;
|
|
164
|
+
let trials=0;
|
|
165
|
+
for(const [at,row] of [...discovery.checks.entries()]){
|
|
166
|
+
if(row.qualification!==sandboxQualification||row.level!=='ready')continue;
|
|
167
|
+
if(!json)console.error(`Running ${row.config.name} inside and outside the sandbox to confirm it...`);
|
|
168
|
+
const hidden=await sandboxTrial(discovery.root,row.config,{hide:true});
|
|
169
|
+
if(!hidden.issue)trials++;
|
|
170
|
+
if(!hidden.issue&&hidden.same){
|
|
171
|
+
row.config={...row.config,qualification:hiddenQualification};row.qualification=hiddenQualification;
|
|
172
|
+
row.reason=row.reason.replace('the whole checkout is its input',"the checkout's files git doesn't ignore are its input")+`; ran both ways at init, exit ${hidden.plain}`;continue;}
|
|
173
|
+
// Without git, nothing is hidden: the whole checkout is tried on its own.
|
|
174
|
+
const trial=hidden.plainRun?await sandboxTrial(discovery.root,row.config,{plain:hidden.plainRun}):
|
|
175
|
+
hidden.issue===hiddenListIssue?await sandboxTrial(discovery.root,row.config):hidden;
|
|
176
|
+
const issue=trialIssue(trial);
|
|
177
|
+
if(!issue){row.reason+=`; ran both ways at init, exit ${trial.plain}; it needs files git ignores (${hidden.issue??trialIssue(hidden)})`;continue;}
|
|
178
|
+
if(row.fallback)discovery.checks[at]={...row.fallback,reason:`Sandboxed recipe unavailable: ${issue}; ${row.fallback.reason}`};
|
|
179
|
+
else{discovery.checks[at]=null;discovery.skipped.push({kind:row.config.name,reason:`${row.config.name}: sandboxed recipe unavailable: ${issue}`});}
|
|
180
|
+
}
|
|
181
|
+
discovery.checks=discovery.checks.filter(Boolean).map(({fallback,...row})=>row);
|
|
182
|
+
discovery.config.checks=discovery.checks.map(row=>row.config);
|
|
133
183
|
const configured=fs.existsSync(configFile),unsupported=Boolean(discovery.observationIssue||
|
|
134
184
|
discovery.manager.issue&&!['npm','yarn','pnpm'].includes(discovery.manager.name)||
|
|
135
185
|
discovery.manager.name==='yarn'&&!discovery.manager.version);
|
|
@@ -144,8 +194,9 @@ if(command==='init'){
|
|
|
144
194
|
workspace:check.workspace||null,level,reason})),config:configFile,
|
|
145
195
|
written:willWrite,existing:configured,writable:!configured&&!unsupported&&discovery.checks.length>0};
|
|
146
196
|
if(!json) {
|
|
147
|
-
console.log(`Node ${summary.node}; ${summary.package_manager||'unknown package manager'}`+
|
|
197
|
+
if(summary.package_manager||!discovery.python)console.log(`Node ${summary.node}; ${summary.package_manager||'unknown package manager'}`+
|
|
148
198
|
`${summary.version?' '+summary.version:''}; ${summary.installation_layout} installation.`);
|
|
199
|
+
if(discovery.python)console.log('Python project; tools run from .venv.');
|
|
149
200
|
if(summary.issue)console.log(`Boundary: ${summary.issue}`);
|
|
150
201
|
for(const row of summary.checks){
|
|
151
202
|
console.log(` ${row.name} (${row.script||row.command?.join(' ')}) — `+
|
|
@@ -167,7 +218,7 @@ if(command==='init'){
|
|
|
167
218
|
if(!samePath(realObservedPath(path.dirname(configFile)),discovery.root))
|
|
168
219
|
error('init writes only at the repository root; use --config there or --dry-run');
|
|
169
220
|
fs.writeFileSync(configFile,JSON.stringify(discovery.config,null,2)+'\n',{flag:'wx'});
|
|
170
|
-
if(!json){console.log(`Created ${configFile}. No checks were executed
|
|
221
|
+
if(!json){console.log(`Created ${configFile}. ${trials?'Sandboxed checks ran both ways to confirm them; no results were recorded.':'No checks were executed.'}`);
|
|
171
222
|
console.log('Ready to qualify means the supported contract will be validated; it is not verification evidence.');
|
|
172
223
|
console.log(`Next: review the config, then redue start and redue run ${JSON.stringify(discovery.checks[0].config.name)}.`);
|
|
173
224
|
console.log('Connect your agent: redue agent setup codex (or claude). Preview with --dry-run.');
|
|
@@ -269,6 +320,7 @@ const checks=publicConfig.checks.map(check=>{
|
|
|
269
320
|
const pnpmTypecheck=check.qualification==='pnpm-tsc-v1';
|
|
270
321
|
const testsDirect=check.qualification===testsQualification;
|
|
271
322
|
const generalScript=check.qualification===scriptQualification;
|
|
323
|
+
const sandboxed=isSandboxed(check.qualification);
|
|
272
324
|
if(check.script&&check.command&&!pnpmTypecheck&&!npmDirect&&!testsDirect)
|
|
273
325
|
error(`${check.name}: choose script or command, not both`);
|
|
274
326
|
const manager=publicConfig.packageManager;
|
|
@@ -278,12 +330,16 @@ const checks=publicConfig.checks.map(check=>{
|
|
|
278
330
|
[`@which:${manager}`,'run',check.script]:null;
|
|
279
331
|
const npmAutomatic=check.qualification==='typescript-noemit-v1';
|
|
280
332
|
const yarnWorkspace=check.qualification==='yarn-workspace-tsc-v1';
|
|
281
|
-
if(check.qualification&&!npmAutomatic&&!npmDirect&&!yarnWorkspace&&!pnpmTypecheck&&!testsDirect&&!generalScript)
|
|
333
|
+
if(check.qualification&&!npmAutomatic&&!npmDirect&&!yarnWorkspace&&!pnpmTypecheck&&!testsDirect&&!generalScript&&!sandboxed)
|
|
282
334
|
error(`${check.name}: unknown qualification contract`);
|
|
283
335
|
// The general recipe runs the root package script through the package manager.
|
|
284
336
|
if(generalScript&&(!['npm','pnpm','yarn'].includes(manager)||!check.script||check.command||check.cwd&&check.cwd!=='.'||
|
|
285
337
|
check.workspace||!scriptKinds.includes(check.kind)))
|
|
286
338
|
error(`${check.name}: the general recipe needs a root typecheck, test or lint script, without command or cwd`);
|
|
339
|
+
// The sandboxed recipe runs a root package script, or a tool from the project's .venv.
|
|
340
|
+
if(sandboxed&&(check.cwd&&check.cwd!=='.'||check.workspace||!sandboxKinds.includes(check.kind)||
|
|
341
|
+
(check.script?!['npm','pnpm','yarn'].includes(manager):!pythonTool(check))))
|
|
342
|
+
error(`${check.name}: the sandboxed recipe needs a root ${sandboxKinds.join(' or ')} script, or a command that runs a tool from .venv/bin, without cwd`);
|
|
287
343
|
if(npmDirect&&(manager!=='npm'||check.cwd&&check.cwd!=='.'||!Array.isArray(check.command)||
|
|
288
344
|
check.command[0]!=='@node'||check.command[1]!=='@typescript-compiler:.'||
|
|
289
345
|
!tscRun({},{script:null,args:check.command.slice(2)}).run?.noEmit))
|
|
@@ -324,7 +380,8 @@ const checks=publicConfig.checks.map(check=>{
|
|
|
324
380
|
check.workspace||'.',...run]]:[];
|
|
325
381
|
const windowsContext=process.platform==='win32'?
|
|
326
382
|
['ComSpec','COMSPEC','PATHEXT','USERPROFILE']:[];
|
|
327
|
-
const autoEnvironment=npmDirect?directEnvironment:testsDirect?testEnvironment:generalScript?scriptEnvironment(manager):
|
|
383
|
+
const autoEnvironment=npmDirect?directEnvironment:testsDirect?testEnvironment:generalScript?scriptEnvironment(manager):
|
|
384
|
+
sandboxed?sandboxEnvironment(Boolean(pythonTool(check)),check.script?manager:null):npmAutomatic?{variables:['NODE_OPTIONS','NODE_PATH','CI','HOME',
|
|
328
385
|
'NODE_ENV','BASH_ENV','ENV',...windowsContext],prefixes:['npm_config_','DYLD_','TSGO_'],
|
|
329
386
|
pathExecutables:['node','npm'],executableIdentity:true}:yarnWorkspace?
|
|
330
387
|
{variables:['NODE_OPTIONS','NODE_PATH','CI','HOME','NODE_ENV','BASH_ENV','ENV',...windowsContext],
|
|
@@ -351,11 +408,18 @@ const checks=publicConfig.checks.map(check=>{
|
|
|
351
408
|
...(npmDirect?{qualificationInterpretation:directInterpretation}:{}),
|
|
352
409
|
...(testsDirect?{qualificationInterpretation:testsInterpretation}:{}),
|
|
353
410
|
...(generalScript?{qualificationInterpretation:scriptInterpretation}:{}),
|
|
411
|
+
...(sandboxed?{qualificationInterpretation:sandboxInterpretationOf(check.qualification)}:{}),
|
|
354
412
|
workspace:check.workspace||null,kind:check.kind||null};
|
|
355
413
|
});
|
|
356
414
|
const runtime={schema:1,provider:'declared-project@1',root,state,checks};
|
|
357
415
|
try{assertOwnedStatePlacement(runtime);}catch(e){error(e.message);}
|
|
358
416
|
const serialized=JSON.stringify(runtime,null,2)+'\n';
|
|
417
|
+
if(command==='stats'){
|
|
418
|
+
const {readStats,renderStats}=await import('../src/stats.mjs');
|
|
419
|
+
const value=readStats(state,publicConfig.checks.map(item=>item.name));
|
|
420
|
+
console.log(json?JSON.stringify(value):renderStats(value));
|
|
421
|
+
await exit(0);
|
|
422
|
+
}
|
|
359
423
|
const maintenance=command==='stop'||command==='remove-state';
|
|
360
424
|
if(maintenance&&!fs.existsSync(runtimeFile)){
|
|
361
425
|
console.log(`No REDUE state exists for ${configFile}.`);
|
|
@@ -395,11 +459,14 @@ const internal=command==='remove-state'?'uninstall':
|
|
|
395
459
|
// Asked when this part of a command runs, after earlier parts may have
|
|
396
460
|
// changed files. REDUE_FORCE=1 runs it anyway.
|
|
397
461
|
if(ifDue&&process.env.REDUE_FORCE!=='1'){
|
|
398
|
-
const {verdict}=await import('../src/claude-hook.mjs');
|
|
462
|
+
const [{verdict},{note,skipNote,dueNote}]=await Promise.all([import('../src/claude-hook.mjs'),import('../src/stats.mjs')]);
|
|
399
463
|
const asked=spawnSync(process.execPath,[fileURLToPath(import.meta.url),'--config',configFile,...(stateOverride?['--state-dir',stateOverride]:[]),
|
|
400
464
|
'status','--sync','--json'],{encoding:'utf8',stdio:['ignore','pipe','ignore'],env:process.env,timeout:600_000});
|
|
401
465
|
let status=null;try{status=JSON.parse(asked.stdout);}catch{}
|
|
466
|
+
const due=dueNote(status,check);
|
|
467
|
+
if(due)note(state,due);
|
|
402
468
|
if(verdict(status,check)){
|
|
469
|
+
note(state,skipNote(status,check,'if-due'));
|
|
403
470
|
const row=status.checks.find(item=>item.name===check);
|
|
404
471
|
if(json)console.log(JSON.stringify({check,skipped:true,result:'PASS',run_id:row.invocation.runId}));
|
|
405
472
|
else console.error(`REDUE: ${check} passed on these exact inputs (run ${row.invocation.runId}); not run again. REDUE_FORCE=1 runs it anyway.`);
|
package/docs/README.md
CHANGED
|
@@ -5,6 +5,7 @@ For users:
|
|
|
5
5
|
- [Install, initialize, run, and remove](user/getting-started.md)
|
|
6
6
|
- [Fixed-build opt-in development pilot](user/pilot.md)
|
|
7
7
|
- [Supported projects and conservative boundaries](user/supported-projects.md)
|
|
8
|
+
- [Sandboxed recipe: typecheck and lint in any language, Python included](user/sandbox.md)
|
|
8
9
|
- [Configuration](user/configuration.md)
|
|
9
10
|
- [Connect coding agents and apply the decision policy](user/agents.md)
|
|
10
11
|
- [JSON contract](user/json.md)
|
package/docs/user/node-script.md
CHANGED
|
@@ -4,10 +4,12 @@ The general recipe records a project's own typecheck or lint script, run
|
|
|
4
4
|
through its package manager, and reports later whether that result still
|
|
5
5
|
applies to the files you have. It is a lower tier than the strict recipes:
|
|
6
6
|
`redue init` tries the [direct TypeScript recipe](direct-typescript.md), the
|
|
7
|
-
pnpm and Yarn TypeScript recipes
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
the
|
|
7
|
+
pnpm and Yarn TypeScript recipes, the [direct test recipe](tests-direct.md)
|
|
8
|
+
and, on macOS and Linux, the [sandboxed recipe](sandbox.md) first. It uses
|
|
9
|
+
the general recipe for a root typecheck or lint script none of them takes,
|
|
10
|
+
and where the sandbox cannot run (at init, or at run time inside Claude
|
|
11
|
+
Code's macOS sandbox). Test scripts use the direct test recipe; `init` does
|
|
12
|
+
not propose the general recipe for them.
|
|
11
13
|
|
|
12
14
|
It supports a project installed by npm, pnpm 9-12 (isolated or hoisted
|
|
13
15
|
`node_modules`) or Yarn with the `node-modules` linker, with one lockfile kind
|
|
@@ -121,8 +123,10 @@ recording-only.
|
|
|
121
123
|
Anything else makes the result UNVERIFIED and names the program: `git`,
|
|
122
124
|
`sleep`, `find`, and native programs from installed packages such as esbuild,
|
|
123
125
|
turbo, tsgo, tsgolint and biome, whose reads, writes and network use REDUE
|
|
124
|
-
cannot see. `init` reads the script the same way
|
|
125
|
-
|
|
126
|
+
cannot see. `init` reads the script the same way. A typecheck or lint script
|
|
127
|
+
gets the [sandboxed recipe](sandbox.md) first where it can run, which covers
|
|
128
|
+
such programs; otherwise a script that starts one stays recording-only, with
|
|
129
|
+
the program in the reason.
|
|
126
130
|
|
|
127
131
|
A command line may use `&&`, `||`, `;`, `|`, quotes without expansions,
|
|
128
132
|
globs in arguments and redirection to `/dev/null`. Variables, command
|
|
@@ -0,0 +1,221 @@
|
|
|
1
|
+
# Sandboxed recipe (typecheck and lint, any language)
|
|
2
|
+
|
|
3
|
+
The sandboxed recipe runs a typecheck or lint command inside an operating
|
|
4
|
+
system sandbox and reports later whether that result still applies to the
|
|
5
|
+
files you have. The sandbox holds for every process the command starts, so it
|
|
6
|
+
covers programs REDUE cannot record: TypeScript 7's native compiler, esbuild,
|
|
7
|
+
oxlint and tsgolint, biome, turbo, and Python tools such as mypy, pyright, ty
|
|
8
|
+
and ruff.
|
|
9
|
+
|
|
10
|
+
**The claim.** A sandboxed result holds for a run that sees only the project's
|
|
11
|
+
files that git doesn't ignore, its installed packages and the toolchain, the
|
|
12
|
+
way CI does. It does not say that a run with your own configuration files, a
|
|
13
|
+
service on your machine or the network would give the same result.
|
|
14
|
+
|
|
15
|
+
A check that needs files git ignores (generated code, another package's
|
|
16
|
+
built declarations) gets the whole checkout instead: its result holds for a
|
|
17
|
+
run that sees the project, git-ignored files included, its installed packages
|
|
18
|
+
and the toolchain. See [files git ignores](#files-git-ignores).
|
|
19
|
+
|
|
20
|
+
## Where it applies
|
|
21
|
+
|
|
22
|
+
- **Node projects:** a root `typecheck` or `lint` script (npm, pnpm, or Yarn
|
|
23
|
+
with the `node-modules` linker). `init` tries the
|
|
24
|
+
[direct TypeScript recipe](direct-typescript.md) and the pnpm and Yarn
|
|
25
|
+
TypeScript recipes first; they stay first where they apply. Then the
|
|
26
|
+
sandboxed recipe, before the [general recipe](node-script.md), since it
|
|
27
|
+
also covers the native programs a script starts. Where init's trials show
|
|
28
|
+
a difference, or the sandbox cannot run, `init` writes the general recipe
|
|
29
|
+
as before. Where the sandbox cannot start at run time (inside Claude Code's
|
|
30
|
+
macOS sandbox), a script the general recipe takes is recorded as it records
|
|
31
|
+
it.
|
|
32
|
+
- **Python projects:** mypy, pyright, ty and ruff installed in the project's
|
|
33
|
+
`.venv` (`uv sync` makes one; Poetry does with `virtualenvs.in-project`).
|
|
34
|
+
`init` uses the arguments your CI workflow, Makefile, tox, nox, justfile or
|
|
35
|
+
shell scripts in `scripts/` run the tool with, through `uv run`, `poetry run`,
|
|
36
|
+
`python -m` or `.venv/bin`. Without them it uses the tool's default:
|
|
37
|
+
`ruff check`, `ty check`, `pyright`, and `mypy` (with `.` when mypy's
|
|
38
|
+
configuration names no `files`).
|
|
39
|
+
- **Platforms:** macOS (Seatbelt, `sandbox-exec`) and Linux 6.7 or later
|
|
40
|
+
(Landlock; the launcher needs Python 3).
|
|
41
|
+
|
|
42
|
+
```sh
|
|
43
|
+
redue init
|
|
44
|
+
redue start
|
|
45
|
+
redue run typecheck
|
|
46
|
+
redue status --sync
|
|
47
|
+
```
|
|
48
|
+
|
|
49
|
+
A Python check in the config:
|
|
50
|
+
|
|
51
|
+
```json
|
|
52
|
+
{"name": "typecheck", "kind": "typecheck", "qualification": "sandbox-v2",
|
|
53
|
+
"command": [".venv/bin/mypy", "src"], "inputs": ["**", "..."], "installedInputs": [".venv/**", "..."]}
|
|
54
|
+
```
|
|
55
|
+
|
|
56
|
+
A Node check names its script, as the general recipe's do:
|
|
57
|
+
`{"name": "lint", "script": "lint", "kind": "lint", "qualification": "sandbox-v2", ...}`.
|
|
58
|
+
`sandbox-v2` hides the files git ignores from the run; `sandbox-v1` runs it on
|
|
59
|
+
the whole checkout.
|
|
60
|
+
|
|
61
|
+
With the Claude Code hook, `uv run mypy src`, `poetry run mypy src`,
|
|
62
|
+
`python -m mypy src`, `.venv/bin/mypy src` and `mypy src` are recognized as
|
|
63
|
+
that check. The last two count only when PATH finds the tool or Python in the
|
|
64
|
+
project's `.venv/bin` first. When the hook runs the check through REDUE, it
|
|
65
|
+
keeps `uv run` or `poetry run` in front, so the environment still syncs first.
|
|
66
|
+
|
|
67
|
+
## Files git ignores
|
|
68
|
+
|
|
69
|
+
`init` runs each check three ways at most: plainly, sandboxed with the files
|
|
70
|
+
git ignores hidden, and, when those two differ, sandboxed with the whole
|
|
71
|
+
checkout (reusing the plain run). The first sandboxed way that gives the same
|
|
72
|
+
result is written: `sandbox-v2` or `sandbox-v1`.
|
|
73
|
+
|
|
74
|
+
With `sandbox-v2`, each run hides what `git ls-files --others --ignored
|
|
75
|
+
--exclude-standard --directory` lists, except `node_modules`, `.venv` and
|
|
76
|
+
runner caches: build output (`dist/`, `.next/`), coverage files, `.env` files.
|
|
77
|
+
Only the project's rules count: its `.gitignore` files and
|
|
78
|
+
`.git/info/exclude`, not your global excludes file. Those files are not
|
|
79
|
+
inputs, so a build or test run that writes them leaves the result CURRENT.
|
|
80
|
+
|
|
81
|
+
When a `sandbox-v2` run fails with a refusal that names a hidden file (a
|
|
82
|
+
missing file counts only when it names one, since Linux tools report an
|
|
83
|
+
unreadable file as missing), REDUE runs the check again in the sandbox
|
|
84
|
+
without hiding anything. That result is kept and compared on every file,
|
|
85
|
+
git-ignored ones included. Run `redue init` again to write `sandbox-v1` for a
|
|
86
|
+
check that keeps needing them.
|
|
87
|
+
|
|
88
|
+
## What the run sees
|
|
89
|
+
|
|
90
|
+
It may read and execute only beneath:
|
|
91
|
+
|
|
92
|
+
- the checkout, without the files git ignores for `sandbox-v2`;
|
|
93
|
+
- the toolchain: Node's installation and the package managers' (with
|
|
94
|
+
Corepack's downloads), or the Python installation the `.venv` was made from;
|
|
95
|
+
- system directories (`/usr`, `/bin`, `/etc` and the like on Linux; `/usr`,
|
|
96
|
+
`/System`, `/Library`, `/opt/homebrew` and the like on macOS);
|
|
97
|
+
- a private home directory and a fresh temporary directory, in REDUE's state.
|
|
98
|
+
|
|
99
|
+
It has no network: no TCP connection, loopback included. On macOS its only
|
|
100
|
+
Unix sockets are in its own temporary directory, which is new for each run.
|
|
101
|
+
On Linux 6.12 and later it cannot reach an abstract Unix socket outside the
|
|
102
|
+
run.
|
|
103
|
+
|
|
104
|
+
What the check leaves running is stopped when it exits, as at the end of a
|
|
105
|
+
CI job: its process group gets SIGTERM, and SIGKILL a second later. On Linux
|
|
106
|
+
the launcher also stops daemons that started their own session. On macOS
|
|
107
|
+
such a daemon outlives the run, but no later run can reach it: its socket
|
|
108
|
+
was in a temporary directory that is gone. The run gets no terminal input.
|
|
109
|
+
|
|
110
|
+
The checkout's `.git` is not an input. Tools still see it, as they do in a CI
|
|
111
|
+
checkout (a linter that applies `.gitignore` only inside a repository does),
|
|
112
|
+
but git commands find no repository: `GIT_DIR` points nowhere. A script that
|
|
113
|
+
runs git (`git diff --exit-code`) gives a different result inside the
|
|
114
|
+
sandbox, and `init` leaves it to the other recipes.
|
|
115
|
+
|
|
116
|
+
It gets a controlled environment, not yours:
|
|
117
|
+
|
|
118
|
+
- the variables the check declares (`TZ`, `LANG`, `CI`, `USER`, `LC_*`,
|
|
119
|
+
`DYLD_*`, `LD_*`, and `NODE_*`, `npm_config_*`, `PNPM_*`, `YARN_*`,
|
|
120
|
+
`COREPACK_*` for Node, or `PYTHON*`, `MYPY*`, `RUFF_*`, `PYRIGHT*`, `TY_*`
|
|
121
|
+
for Python), with your values;
|
|
122
|
+
- your PATH, without the project's `.venv/bin`, which a Python check gets
|
|
123
|
+
first instead, so `uv run` and an activated environment give the same PATH;
|
|
124
|
+
- `HOME`, `TMPDIR` and the XDG directories in its private directories;
|
|
125
|
+
`VIRTUAL_ENV` for a Python check; Corepack's home where Corepack keeps it.
|
|
126
|
+
|
|
127
|
+
## What REDUE claims
|
|
128
|
+
|
|
129
|
+
CURRENT means the last sandboxed run's inputs are unchanged:
|
|
130
|
+
|
|
131
|
+
- every file in the checkout git doesn't ignore (`sandbox-v2`), or every file
|
|
132
|
+
in it, git-ignored files and git-ignored links to files in it included
|
|
133
|
+
(`sandbox-v1`), except `.git` and runner caches (the general recipe's, plus
|
|
134
|
+
`__pycache__`, `.mypy_cache`, `.ruff_cache` and `.pytest_cache`);
|
|
135
|
+
- `node_modules` and `.venv`, observed as installed files, by identity;
|
|
136
|
+
- the environment the run gets (without display variables such as `TERM`
|
|
137
|
+
and `COLUMNS`), and the places it may read;
|
|
138
|
+
- Node's and the package managers' executables, or the interpreter the
|
|
139
|
+
`.venv` was made from, by file identity;
|
|
140
|
+
- every directory on PATH, by identity: installing, removing or upgrading a
|
|
141
|
+
program in one makes the result STALE;
|
|
142
|
+
- `.git/info/exclude` and `.git/config`, by content: ruff and biome read the
|
|
143
|
+
ignore rules there, and they decide what `sandbox-v2` hides. For a project
|
|
144
|
+
in a subdirectory of its repository, also the `.gitignore` files above it.
|
|
145
|
+
|
|
146
|
+
## Telling a refusal from a real failure
|
|
147
|
+
|
|
148
|
+
A run the sandbox refused something fails like any other. REDUE does not run
|
|
149
|
+
every failure twice to find out:
|
|
150
|
+
|
|
151
|
+
- `init` runs each proposed check twice, plainly and sandboxed, and proposes
|
|
152
|
+
it only when both give the same exit status and the same output, and a
|
|
153
|
+
failure shows no sign of a refusal. The output is compared without colors,
|
|
154
|
+
timings, the temporary and home directories, and lines about caches.
|
|
155
|
+
Otherwise `init` writes what it would have without this recipe and says why.
|
|
156
|
+
These are the only runs `init` makes.
|
|
157
|
+
- When a `sandbox-v2` run fails with a refusal that names a hidden file, it
|
|
158
|
+
runs again in the sandbox without hiding (see
|
|
159
|
+
[files git ignores](#files-git-ignores)).
|
|
160
|
+
- When a sandboxed run fails and its output has a line that looks like a
|
|
161
|
+
refusal ("Operation not permitted", "Permission denied", "Read-only file
|
|
162
|
+
system", a name lookup or a connection that failed), REDUE runs the check
|
|
163
|
+
again outside the sandbox. If that exits differently, the result is the
|
|
164
|
+
plain run's and UNVERIFIED: "the sandbox changed the result", with the
|
|
165
|
+
line. A failure without such a line is not run again.
|
|
166
|
+
|
|
167
|
+
## The private home
|
|
168
|
+
|
|
169
|
+
The run's home directory is kept between runs, per check, so caches tools
|
|
170
|
+
build there (font caches, a type checker's cache) are not rebuilt every run.
|
|
171
|
+
It starts empty: your configuration files are not in it. What a run leaves
|
|
172
|
+
there can change a later run, and REDUE trusts it like a runner cache. Remove
|
|
173
|
+
it with `redue remove-state`.
|
|
174
|
+
|
|
175
|
+
## Limits
|
|
176
|
+
|
|
177
|
+
- Inside Claude Code's own sandbox on macOS, `sandbox-exec` cannot start: the
|
|
178
|
+
run goes ahead unsandboxed and is UNVERIFIED ("the run was not sandboxed").
|
|
179
|
+
On Linux, Landlock works inside it.
|
|
180
|
+
- Windows is not supported. The check stays recording-only there.
|
|
181
|
+
- Linux before 6.7 (Landlock ABI below 4) cannot refuse TCP connections, so a
|
|
182
|
+
run there is UNVERIFIED. UDP is not refused on Linux.
|
|
183
|
+
- System files (time zones, certificates, `/etc`) are not inputs: a change to
|
|
184
|
+
them does not make a result STALE.
|
|
185
|
+
- A program that reads files in `.git` itself, without git (a commit hash for
|
|
186
|
+
a version string), reads something REDUE does not observe: a commit does
|
|
187
|
+
not make the result STALE.
|
|
188
|
+
- Runner caches in the checkout are trusted as the general recipe trusts them.
|
|
189
|
+
- On Linux, a run can still reach a daemon started outside it through a Unix
|
|
190
|
+
socket file (Landlock does not cover them), and what that daemon reads is
|
|
191
|
+
not seen. On macOS it cannot.
|
|
192
|
+
- On Linux, uv hard-links `.venv` files to its cache. The cache is
|
|
193
|
+
toolchain: a file rewritten through the cache, not through the `.venv`, is
|
|
194
|
+
noticed only when the observer next reads the `.venv` (when it starts).
|
|
195
|
+
- A `.venv` link that leads out of the checkout, other than the interpreter's,
|
|
196
|
+
keeps the check recording-only and names the link, and so does a `.venv`
|
|
197
|
+
that uses the system site-packages.
|
|
198
|
+
- A git-ignored link to a directory, or out of the checkout, is not observed,
|
|
199
|
+
and neither is a `node_modules` directory outside the installed ones (a
|
|
200
|
+
nested example project's).
|
|
201
|
+
- A script that changes files (`--fix`, `--write`) does not get this recipe:
|
|
202
|
+
`init` would run it twice.
|
|
203
|
+
- With `sandbox-v2`, a run cannot read a hidden file, but it can still see
|
|
204
|
+
that one exists. On macOS it sees the names of hidden files and folders at
|
|
205
|
+
the level where they start (`dist` in the checkout's root). On Linux it can
|
|
206
|
+
also list a hidden folder and see its files' sizes and times: Landlock
|
|
207
|
+
restricts neither beneath a folder the run may list. A tool that acts on
|
|
208
|
+
whether a git-ignored file exists, without reading it, is not covered: a
|
|
209
|
+
new or removed one does not make the result STALE. Typecheckers and
|
|
210
|
+
linters read what they act on; a test that skips itself when a built file
|
|
211
|
+
is missing does not, which is why hidden files are not used for tests yet.
|
|
212
|
+
- On Linux, a file a `sandbox-v2` run creates in a folder that holds hidden
|
|
213
|
+
files (the checkout's root, when anything in it is hidden) cannot be read
|
|
214
|
+
back in the same run. `init`'s trial and the rerun without hiding catch a
|
|
215
|
+
check that needs to.
|
|
216
|
+
- A file whose git status changes without a change to its content (`git add
|
|
217
|
+
-f` of an ignored file, `git rm --cached` of a tracked one) is noticed only
|
|
218
|
+
when its content next changes. A change to `.gitignore` or
|
|
219
|
+
`.git/info/exclude` makes the result STALE.
|
|
220
|
+
- A run that moves a hidden folder into a visible place can read it there.
|
|
221
|
+
That changes the checkout during the run, so the run is not reused.
|
|
@@ -18,11 +18,15 @@ build scripts. It prefers familiar script names (for typecheck: `typecheck`,
|
|
|
18
18
|
(`typecheck:test`, `tsc:build`; not a watch script), then unique recognizable
|
|
19
19
|
`tsc`, Vitest/Jest, ESLint, or Vite commands under other names. Multiple
|
|
20
20
|
matches are reported as ambiguous rather than guessed. It never creates a new
|
|
21
|
-
project script. Without `--recipe`, `init` tries the strict recipes below first,
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
|
|
21
|
+
project script. Without `--recipe`, `init` tries the strict recipes below first, then,
|
|
22
|
+
for a root typecheck or lint script, the [sandboxed recipe](sandbox.md) on
|
|
23
|
+
macOS and Linux (the script run inside an OS sandbox, with the files git
|
|
24
|
+
ignores hidden when it gives the same result that way, else on the whole
|
|
25
|
+
checkout), then the [general recipe](node-script.md): the script run
|
|
26
|
+
through the package manager with every Node process recorded and the whole
|
|
27
|
+
checkout as its input. Python projects get the sandboxed recipe for mypy,
|
|
28
|
+
pyright, ty and ruff in their `.venv`. `--recipe typescript-direct`,
|
|
29
|
+
`tests-direct`, `node-script` or `sandbox` selects one recipe alone. The direct TypeScript recipe supports
|
|
26
30
|
a typecheck script of `tsc --noEmit` runs in any argument order, chained runs,
|
|
27
31
|
or no typecheck script at all (a proposed `tsc --noEmit`; see
|
|
28
32
|
[script forms](direct-typescript.md#script-forms)), in a single-package npm installation using
|