Commit 66e76a9
fix(ci): make the dogfood gate .deed-aware (#379)
## Why
This repo's Dogfood Gate enforced **A2ML, a format that no longer
exists.** A2ML was
retired after a name clash with the ML community; the live s-expression
format is
**`.deed`**, which has a grammar (`DEED-GRAMMAR-SPEC`) and is bound for
IANA.
There are **two** defects here, and fixing only the first would have
been a fake cure.
### 1. The validator was frozen before the rename
`.githooks/validate-a2ml.sh` was vendored at 13,737 bytes by commit
`d5f229d`
(2026-07-27, *"ci: vendor validation scripts and remove remote action
pins"*) — six weeks
**before** the 09-03 DEED rename. It contains **zero** `.deed`
references.
Replaced with the canonical dual-accept body from
`hyperpolymath/deed-ecosystem/validate-action` @ `f9d999b6` (17,544 b),
which dispatches on
the DEED s-expression head —
`(estate-deed|repo-deed|estate-atlas-deed|praxis-deed)` on the
first form, per `DEED-GRAMMAR-SPEC <<concrete-syntax>>` — and **keeps
`.a2ml` passing as
legacy**. This repo's own `SPDX-License-Identifier` line is preserved,
not overwritten.
### 2. 🚨 The step that *gates* that validator counted only `*.a2ml`
```yaml
COUNT=$(find . -name '*.a2ml' -not -path './.git/*' | wc -l)
- name: Validate A2ML manifests
if: steps.detect.outputs.count > 0 # <- the actual gate
run: bash .githooks/validate-a2ml.sh
```
On a repo that had migrated to `.deed`, `COUNT` would be **0**, the
validate step would be
**skipped**, and the job would report **green having validated
nothing**. Upgrading the
validator alone would have been invisible.
The selector now admits both extensions — **the same shape the K9 job in
this very file
already uses** (`find . \( -name '*.k9' -o -name '*.k9.ncl' \)`).
### 3. The gate was instructing maintainers to author the dead format
The warning and summary told every RSR repo to create
`0-AI-MANIFEST.a2ml` and to run
`a2mliser init`. Reworded to name `.deed` as the format to author,
`.a2ml` as legacy.
## Measured, one denominator
Census across the **75** repos carrying a vendored
`.githooks/validate-a2ml.sh`
(75 unique repos; an earlier figure of 79 was a *row* count — 4 slugs
were duplicated by
two local checkouts reaching one remote):
| property | count |
|---|---|
| detect step counts only `*.a2ml` | **75 / 75** |
| workflow mentions `deed` at all | **0 / 75** |
| carry any `.deed` file at `origin/main` | **0 / 75** |
| validator at 13,737 b / md5 `8f61f8da` / MPL-2.0 | 74 |
| `echidna` — 13,747 b / `c1769fc4` / **AGPL-3.0-or-later** | 1 |
## Why edit 2 is safe
**It is behaviour-neutral today.** `find A -o B` is a strict superset of
`find A`, and
**0 of 75 repos carry a `.deed` file**, so `COUNT` is unchanged on every
repo in the
population. The change is purely forward-correct: it stops the gate
going silently green
the day a repo migrates.
## Verification
- `bash -n` on the new validator: clean.
- `python3 -c 'yaml.safe_load(...)'` on the edited workflow: parses.
- The sweep **asserts its own precondition** per repo — exact size
**and** md5 **and** SPDX
before any write — and refuses (skips + flags) on any mismatch rather
than overwriting an
unrecognised file. `echidna` is handled by the same assert, which is
what prevented a
blanket copy from silently relicensing it AGPL-3.0-or-later → MPL-2.0.
- Post-edit asserts: the restored SPDX line is present, the widened
selector is present, and
**no a2ml-only `find` survived**.
- Upstream regression evidence: the 17,544-byte body is already merged
on `origin/main` in
both `a2ml-ecosystem` and `deed-ecosystem`; 14 repos / 725 `.a2ml` files
produced
identical exit codes under old and new validators, with a planted
positive and negative.
## Not changed here
- `strict` stays `false`. Flipping it is a separate, measured decision:
8 of 10 sampled
repos go rc=0 → rc=1.
- No `.a2ml` file is renamed. Extension migration is explicitly held
pending a dry-run
manifest (19,477 tracked `.a2ml` vs 15 `.deed`).
- The canonical validator's own line 5 still reads
`# validate-a2ml.sh — A2ML manifest validation script`. Left
byte-identical to upstream
on purpose, so *"is this repo at canonical?"* stays answerable by md5.
**Flagged for an
upstream fix in `deed-ecosystem`**, where it corrects once for every
consumer.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
https://claude.ai/code/session_0168Bgpez8mFBcAqYAj8VgEx
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>1 parent 6e9cacf commit 66e76a9
2 files changed
Lines changed: 95 additions & 32 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
4 | 4 | | |
5 | 5 | | |
6 | 6 | | |
7 | | - | |
| 7 | + | |
8 | 8 | | |
9 | 9 | | |
10 | 10 | | |
| |||
89 | 89 | | |
90 | 90 | | |
91 | 91 | | |
92 | | - | |
| 92 | + | |
93 | 93 | | |
94 | 94 | | |
95 | 95 | | |
| |||
123 | 123 | | |
124 | 124 | | |
125 | 125 | | |
| 126 | + | |
126 | 127 | | |
127 | 128 | | |
128 | 129 | | |
| |||
150 | 151 | | |
151 | 152 | | |
152 | 153 | | |
| 154 | + | |
| 155 | + | |
| 156 | + | |
| 157 | + | |
| 158 | + | |
| 159 | + | |
| 160 | + | |
| 161 | + | |
| 162 | + | |
| 163 | + | |
| 164 | + | |
| 165 | + | |
| 166 | + | |
| 167 | + | |
| 168 | + | |
| 169 | + | |
| 170 | + | |
| 171 | + | |
| 172 | + | |
| 173 | + | |
| 174 | + | |
| 175 | + | |
| 176 | + | |
153 | 177 | | |
154 | 178 | | |
155 | 179 | | |
| |||
162 | 186 | | |
163 | 187 | | |
164 | 188 | | |
| 189 | + | |
| 190 | + | |
| 191 | + | |
| 192 | + | |
| 193 | + | |
| 194 | + | |
| 195 | + | |
| 196 | + | |
| 197 | + | |
| 198 | + | |
| 199 | + | |
| 200 | + | |
| 201 | + | |
165 | 202 | | |
166 | 203 | | |
167 | 204 | | |
168 | 205 | | |
169 | 206 | | |
170 | 207 | | |
171 | 208 | | |
172 | | - | |
| 209 | + | |
| 210 | + | |
| 211 | + | |
| 212 | + | |
| 213 | + | |
173 | 214 | | |
174 | 215 | | |
175 | | - | |
| 216 | + | |
176 | 217 | | |
177 | 218 | | |
178 | 219 | | |
| |||
203 | 244 | | |
204 | 245 | | |
205 | 246 | | |
206 | | - | |
207 | | - | |
208 | | - | |
209 | | - | |
210 | | - | |
211 | | - | |
212 | | - | |
213 | | - | |
214 | | - | |
215 | | - | |
| 247 | + | |
| 248 | + | |
| 249 | + | |
| 250 | + | |
| 251 | + | |
| 252 | + | |
| 253 | + | |
| 254 | + | |
| 255 | + | |
| 256 | + | |
| 257 | + | |
| 258 | + | |
| 259 | + | |
| 260 | + | |
| 261 | + | |
| 262 | + | |
| 263 | + | |
| 264 | + | |
| 265 | + | |
| 266 | + | |
| 267 | + | |
216 | 268 | | |
217 | | - | |
218 | | - | |
219 | | - | |
| 269 | + | |
| 270 | + | |
| 271 | + | |
| 272 | + | |
| 273 | + | |
| 274 | + | |
| 275 | + | |
| 276 | + | |
| 277 | + | |
220 | 278 | | |
221 | 279 | | |
222 | 280 | | |
223 | 281 | | |
224 | 282 | | |
225 | 283 | | |
226 | | - | |
| 284 | + | |
| 285 | + | |
| 286 | + | |
| 287 | + | |
| 288 | + | |
| 289 | + | |
227 | 290 | | |
228 | 291 | | |
229 | 292 | | |
| |||
259 | 322 | | |
260 | 323 | | |
261 | 324 | | |
262 | | - | |
| 325 | + | |
263 | 326 | | |
264 | 327 | | |
265 | 328 | | |
| |||
281 | 344 | | |
282 | 345 | | |
283 | 346 | | |
284 | | - | |
| 347 | + | |
285 | 348 | | |
286 | 349 | | |
287 | 350 | | |
288 | | - | |
| 351 | + | |
289 | 352 | | |
290 | 353 | | |
291 | | - | |
292 | | - | |
| 354 | + | |
| 355 | + | |
293 | 356 | | |
294 | 357 | | |
295 | 358 | | |
| |||
307 | 370 | | |
308 | 371 | | |
309 | 372 | | |
310 | | - | |
| 373 | + | |
311 | 374 | | |
312 | 375 | | |
313 | 376 | | |
314 | 377 | | |
315 | 378 | | |
316 | 379 | | |
317 | 380 | | |
318 | | - | |
| 381 | + | |
319 | 382 | | |
320 | 383 | | |
321 | 384 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
40 | 40 | | |
41 | 41 | | |
42 | 42 | | |
43 | | - | |
| 43 | + | |
44 | 44 | | |
45 | 45 | | |
46 | | - | |
| 46 | + | |
47 | 47 | | |
48 | 48 | | |
49 | 49 | | |
| |||
56 | 56 | | |
57 | 57 | | |
58 | 58 | | |
59 | | - | |
| 59 | + | |
60 | 60 | | |
61 | | - | |
| 61 | + | |
62 | 62 | | |
63 | 63 | | |
64 | 64 | | |
65 | 65 | | |
66 | | - | |
| 66 | + | |
67 | 67 | | |
68 | 68 | | |
69 | 69 | | |
| |||
326 | 326 | | |
327 | 327 | | |
328 | 328 | | |
329 | | - | |
| 329 | + | |
330 | 330 | | |
331 | 331 | | |
332 | 332 | | |
| |||
380 | 380 | | |
381 | 381 | | |
382 | 382 | | |
383 | | - | |
| 383 | + | |
384 | 384 | | |
385 | 385 | | |
386 | 386 | | |
| |||
0 commit comments