profiles: delete grind-free Fast/Slim; Fast128/Slim128 take their names (proof-IO v22) - #39
Merged
Merged
Conversation
…es (proof-IO v22) Phase 2 of the bloat plan, item 0 (ledger §C, Ron's call 2026-08-27). The grind-free `Fast`/`Slim` with the +1/level ladder are gone. The `Fast128`/`Slim128` schedules — aggressive +2/level ladder, 16-bit query PoW at every level (query term 112 bits + PoW, work-normalized to 128), larger deep-level batch grinding — now carry the `Fast`/`Slim` names, so every selector and the enum Default get them without moving. The 28 `m*_fast.toml`/`m*_slim.toml` are replaced by the renamed `*128` files (98 -> 70 embedded configs); `gen_ligerito_configs` regenerates the new set byte-identically, so the derivation and the embedded files agree. `Fast100`/`Slim100`/`Secure` are untouched; the *100 twins are documented as frozen historical schedules rather than "Fast with one knob changed". Transcript-moving for strict Fast/Slim users: proof-IO VERSION 21 -> 22 with a changelog note. Re-pinned, two deterministic print runs each: union_m6_fixtures (6 digests), union_element (7), transcript_shape (1). Tower tape pins hold — Chain128 already selected the *128 twins. Verified: full workspace suite (release, --no-fail-fast) green after the re-pins; clippy -D warnings clean; the CUDA host-only ladder replay (make ligerito_f256_host, m22 fast) matches the F256 driver on every proof field. Docs: enum docstrings, ligerito.rs schedule comments, recursion-100-128-variants.md, a dated addendum in the 128-bit audit (its "Fast lambda_query = 0" rows describe the deleted profile), the plan's Phase 2 list and the ledger's execution note. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
rothblum
added a commit
that referenced
this pull request
Aug 31, 2026
retire Keccak and the Merkle-path product (stacked on #39)
rothblum
added a commit
that referenced
this pull request
Aug 31, 2026
…se-0 refactor Brings the branch up to the post-stack main (8a36c91). Resolutions: - keccak.rs, keccak3.rs, merkle_path.rs, merkle_path_common.rs: main deleted them (Phase 2 retirements); this branch's edits to them were mechanical (imports, the VerifyError rename), so the deletions win. - mixed.rs / prover.rs / merkle_r1cs.rs: main's side (the Merkle registry tiers, the timed-prove chain and the walker are gone); the unused std::time::Instant import went with it. - transcript_record.rs: this branch's re-export of flock-transcript wins; main's TranscriptOp::carries_payload() is ported into flock-transcript/src/transcript_record.rs. - proof_io.rs: this branch's trimmed doc style wins, but the version is main's v22 with a short note on what v22 means (profile consolidation, retired registry codes 3/4); the long per-version history stays in git. - test_rng moved DOWN into flock-field (flock-core re-exports it): git's rename detection carried main's test-RNG edits into the moved field files, and flock-field cannot depend on flock-core. Every flock_core::test_rng::Rng path still works. - Cargo.lock: regenerated by the build. Verified: cargo check + clippy -D warnings clean across the workspace (aarch64); full suite + tower pins run follows. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Phase 2, item 0 of the bloat plan (ledger §C): the profile consolidation. The grind-free
Fast/Slim(+1 rate/level ladder) are deleted, and theFast128/Slim128schedules take their names: aggressive +2/level ladder, 16-bit query PoW at every level (query term 112 bits + PoW, work-normalized to 128), larger deep-level batch grinding. Every selector and the enumDefaultkeep theFast/Slimnames, so no call site moves.Fast100/Slim100/Secureare untouched; the*100twins are documented as frozen historical schedules.*128files take their place (98 → 70 embedded configs). Runninggen_ligerito_configsregenerates the new set byte-identically, so the derivation and the embedded files agree.VERSION21 → 22 with a changelog note: no payload field changes, but a v21fast/slimproof carries a different query/rate/PoW schedule than this build derives for those names.union_m6_fixtures(6 digests),union_element(7),transcript_shape(1). Tower tape pins hold:Chain128already selected the*128twins.recursion-100-128-variants.md, a dated addendum at the top of the 128-bit audit (its "Fast useslambda_query = 0" rows describe the deleted profile), the plan's Phase 2 list, the ledger's execution note.Motivation
Two twins per strict profile doubled the config matrix and the grinding-policy matches, and the review of #37 found the ledger's premise for the merge was wrong (the twins did not differ "only in the rate ladder"). Ron's decision on 2026-08-27: keep the
*128schedules, PoW included, under the base names.Validation
Full workspace suite (release,
--no-fail-fast) green after the re-pins. The ignored tower suite (12 tests, tape pins, spine convergence, m32 headline) green. Clippy-D warningsclean on aarch64 and on the exact CI x86_64 lint leg. The CUDA host-only ladder replay (make ligerito_f256_host, m22 fast) matches the F256 driver on every proof field with the new schedule. The Blackwell GPU CI job still needssuccinct-gpu-05back; the CUDA kernels take the ladder, per-level rates and grinding from the dumped config, so no kernel change was needed.Caveats
Transcript-moving for strict
Fast/Slimusers. Anyone holding v21 proofs under those profiles must re-prove;fast100/slim100/secureproofs are unaffected apart from the version byte.🤖 Generated with Claude Code