1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
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
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
//! Substrate primitive for BLAKE3 hex digests.
//!
//! Two peer entries — [`hex_blake3`] for the in-memory-buffer shape
//! every three-pillar attestation producer that fed BLAKE3 a single
//! buffer restated by hand pre-lift, and [`hex_blake3_hash`] for the
//! streaming-digest shape (`hex::encode(hash.as_bytes())` where
//! `hash = blake3::Hasher::finalize()`) every consumer that folded
//! per-item updates into a `Hasher` before finalizing walked.
//!
//! # Crate-layer partition on the byte-input axis
//!
//! The byte-input peer [`hex_blake3`] delegates through
//! [`tatara_lisp::hash::hex_blake3_of_bytes`] — the workspace-wide
//! byte-input owner opened at the `tatara-lisp` layer (commit
//! `524e543`, sibling of `hex_blake3_of_json`). Both byte-input
//! owners now share ONE canonical spelling via that substrate:
//! `blake3::hash(bytes).to_hex().to_string()`. A future encoding
//! change (base32, base64url, uppercase hex for a downstream tool)
//! at [`tatara_lisp::hash::hex_blake3_of_bytes`] reaches every
//! byte-input consumer across the workspace — the three
//! `tatara-closed-loop-probe` receipt pillars, the `tatara-reconciler`
//! `render`/`phase_machine` composers, `tatara-export-worker`'s
//! event-run digest, `crate::hostname::short_hex_blake3`, and every
//! future consumer — through ONE edit. Pre-lift the workspace had
//! two independent byte-input owners (`tatara-process::hash::hex_blake3`
//! keyed off `hex::encode(<hash>.as_bytes())`, and
//! `tatara-lisp::hash::hex_blake3_of_bytes` keyed off
//! `.to_hex().to_string()`); post-lift only the streaming-hash peer
//! [`hex_blake3_hash`] (structurally distinct, receives `&blake3::Hash`
//! rather than `&[u8]`) keeps the `hex::encode(<hash>.as_bytes())`
//! spelling — a spelling required by consumers that already hold a
//! finalized `blake3::Hash`, not a byte buffer.
//!
//! Return type is `String` for wire-shape stability with the pre-lift
//! consumers — `ReceiptEnvelope.{intent,artifact,control}_hash` are
//! typed as `String`, so a `Cow`/`&str` return would force allocation
//! at every call site regardless.
//!
//! # Which peer to call
//!
//! - Have `&[u8]` in hand → [`hex_blake3`]. Delegates to
//! [`tatara_lisp::hash::hex_blake3_of_bytes`], the workspace-wide
//! byte-input owner.
//! - Have a `blake3::Hasher` you already folded per-item updates into
//! → `hex_blake3_hash(&h.finalize())`. Skips the one-shot round-trip
//! through `&[u8]` that would force the caller to materialize the
//! full input buffer just to re-hash it.
/// Compute the lowercase 64-char BLAKE3 hex digest of `bytes`.
///
/// # Invariants
///
/// - **Length:** the returned string is always exactly 64 chars
/// (BLAKE3's 32-byte digest encoded as lowercase hex).
/// - **Charset:** every char is one of `[0-9a-f]` (lowercase).
/// - **Determinism:** byte-identical output across runs for the same
/// input; matches both the `hex::encode(blake3::hash(x).as_bytes())`
/// and `blake3::hash(x).to_hex().to_string()` pre-lift spellings.
///
/// # `#[must_use]`
///
/// Every consumer either stores the returned hex into a receipt
/// pillar (`intent_hash`, `artifact_hash`, `control_hash`) or feeds
/// it into a DNS slot / stable name. Dropping the return means the
/// hash was computed for no observable reason — the attribute
/// surfaces that as a warning at every call site.
///
/// # Delegation
///
/// Routes through [`tatara_lisp::hash::hex_blake3_of_bytes`] — the
/// workspace-wide byte-input owner (commit `524e543` opened it at the
/// `tatara-lisp` layer as the sibling of `hex_blake3_of_json` on the
/// value-input axis). Pre-lift the workspace had TWO independent
/// byte-input owners: this crate's `hex_blake3` keyed off
/// `hex::encode(<hash>.as_bytes())`, and `tatara-lisp`'s
/// `hex_blake3_of_bytes` keyed off `.to_hex().to_string()`. Both
/// produced byte-identical output but through two spellings, so a
/// future encoding change would have to land at both sites or
/// silently break receipt / stable-name parity at the corner that
/// missed the update. Post-lift the byte-input axis has ONE canonical
/// owner, and BOTH this crate's peer and `tatara-lisp`'s peer share
/// its spelling by construction — a future re-encoding (base32,
/// base64url, uppercase, an alternate `to_hex` shape on a future
/// blake3 major-version bump) lands at ONE substrate function and
/// reaches every downstream three-pillar / receipt / hostname /
/// export-run identity slot mechanically.
///
/// The streaming peer [`hex_blake3_hash`] cannot fold into the same
/// owner because it takes `&blake3::Hash` rather than `&[u8]` — a
/// structurally distinct entry point for consumers that folded
/// per-item updates into a `Hasher` before finalizing. Both peers
/// still agree byte-for-byte (pinned at
/// [`tests::hex_blake3_bytes_form_delegates_through_hex_blake3_hash`]
/// on the two-corner axis, and at
/// [`tests::hex_blake3_delegates_through_tatara_lisp_hex_blake3_of_bytes`]
/// on the cross-crate axis).
/// Streaming-digest peer of [`hex_blake3`] — the ONE substrate owner
/// of the 1-link `hex::encode(hash.as_bytes())` encoding step every
/// three-pillar producer that folded per-item updates into a
/// `blake3::Hasher` walked pre-lift.
///
/// # Why it exists
///
/// The `Hasher::finalize() → hex::encode(<Hash>.as_bytes())` chain was
/// hand-authored at TWO sites past the ★★ PRIME-DIRECTIVE ≥ 2
/// duplication threshold, each carrying a `Hasher` it fed per-item
/// updates into before finalizing:
///
/// * [`crate::three_pillar::compose_root`] — folds the domain tag +
/// the four pillars (artifact, control, intent, previous) into a
/// `blake3::Hasher`, then encodes the final hash. Pre-lift ended
/// with `hex::encode(h.finalize().as_bytes())` inline; post-lift
/// ends with `hex_blake3_hash(&h.finalize())`.
/// * `tatara-reconciler::phase_machine::handle_running` — folds each
/// observed FluxCD resource's `(apiVersion, kind, ns, name)`
/// 4-slot identity into a `blake3::Hasher`, then encodes the final
/// hash as the artifact-pillar input for the ATTEST step.
///
/// Pre-lift the one-shot corner ([`hex_blake3`]) and the streaming
/// corner both restated `hex::encode(...as_bytes())` at their own
/// bodies, leaving a two-place drift trap — a future swap onto
/// `blake3::Hash::to_hex().to_string()` (a subtler pre-existing
/// spelling that appears at `crate::hostname::short_hex_blake3`), an
/// uppercase-hex flip for a downstream tool, or a base32-encoded
/// variant would have to land at BOTH sites or silently break receipt
/// verification at the corner that missed the update. Post-lift the
/// terminal hex step lives at ONE substrate owner and the one-shot
/// peer's body reduces to `hex_blake3_hash(&blake3::hash(bytes))`.
///
/// # Invariants
///
/// Same shape as [`hex_blake3`]: 64 chars of lowercase hex, every
/// char in `[0-9a-f]`, byte-identical to the pre-lift `hex::encode(<
/// hash>.as_bytes())` spelling.
///
/// # `#[must_use]`
///
/// Every consumer either stores the returned hex into a receipt
/// pillar or a `composed_root` slot. Dropping the return means the
/// caller finalized a `Hasher` for no observable reason — the
/// attribute surfaces that as a warning at every call site.
///
/// Theory anchor: THEORY.md §VI.1 (generation over composition — the
/// `hex::encode(...as_bytes())` step recurred at two hand-authored
/// sites past the ★★ PRIME-DIRECTIVE ≥ 2 duplication trigger, and is
/// lifted to ONE owner here). THEORY.md §II.1 invariant 5
/// (composition preserves proofs — the pin
/// [`tests::hex_blake3_hash_matches_pre_lift_hex_encode_spelling_bytewise`]
/// binds the streaming corner byte-identically to the pre-lift
/// spelling, and the cross-corner pin
/// [`tests::hex_blake3_bytes_form_delegates_through_hex_blake3_hash`]
/// binds the one-shot peer to `hex_blake3_hash` so a regression in
/// either surfaces at ONE substrate pin rather than as silent
/// composed_root drift across every downstream consumer).