@@ -228,6 +228,47 @@ Phase 3:
228228- Release-gate hardening for packaged evidence, reserved feature placeholders,
229229 unsafe/SIMD boundary validation, and cross-target SIMD feature bundles.
230230
231+ ### v0.6
232+
233+ - Expand Kani proof coverage for length helpers, in-place decode bounds, and
234+ selected scalar decoder invariants when Kani supports the pinned Rust
235+ toolchain.
236+ - Strengthen Miri coverage for scalar, in-place, clear-tail, and stream state
237+ machine behavior that Miri can execute practically.
238+ - Add a documented constant-time verification plan for the ` ct ` module,
239+ including generated-code review requirements before making cryptographic
240+ constant-time claims.
241+ - Stabilize fuzz corpus handling and document which corpus inputs are accepted
242+ into the repository.
243+
244+ ### v0.7
245+
246+ - Replace inactive SIMD encode prototypes with real candidate implementations
247+ only when scalar differential tests, fuzz evidence, target-feature checks,
248+ unsafe inventory updates, and benchmark evidence are complete.
249+ - Keep scalar as the default fallback and require runtime backend policy tests
250+ for every admitted accelerated backend.
251+ - Publish per-architecture benchmark evidence for any performance claim,
252+ including CPU, OS, Rust version, command, and raw output.
253+
254+ ### v0.8
255+
256+ - Consider admitting async streaming wrappers only after ` docs/ASYNC.md `
257+ requirements are met, including dependency review, cancellation behavior,
258+ drop cleanup behavior, chunk-boundary tests, and release evidence with the
259+ async feature enabled.
260+ - If async remains unjustified, keep ` tokio ` inert and spend this milestone on
261+ stream ergonomics, documentation, and additional framed-protocol tests.
262+
263+ ### v0.9
264+
265+ - Release-candidate hardening for ` v1.0 ` : API audit, documentation audit,
266+ migration guide refresh, fuzz corpus review, benchmark review, and release
267+ evidence rehearsal.
268+ - Decide whether constant-time decode is formally guaranteed with supporting
269+ evidence or explicitly documented as constant-time-oriented only.
270+ - Freeze dependency policy and feature admission rules for ` v1.0 ` .
271+
231272### v1.0
232273
233274- Kani proofs complete for scalar in-place decode.
0 commit comments