Repository navigation
Merge PFT proof replay into master - #18
Merged
Merged
Conversation
Mostly corresponds to CakeML's Word8Array.
In particular before indexing into arrays or calling Candle functions which could fail. That way, we have an easier time pinpointing where the failure happened.
Some of the issues were due to HOL Light shadowing kernel definitions such as new_specification. We avoid this issue by using qualified names. pft.ml is essentially a prefix of hol_lib.ml that does not include any (HOL) definitions. This allows us to replay proof traces that define constants such as bool's T.
As discussed on Zulip, the spec is updated and the variables will appear first in the tuple list representing substitutions.
Assisted-by: Claude:claude-opus-4-6
Assisted-by: Claude:claude-opus-4-6
Replay works. At the end we run into an error, since it is trying to interpret the footer as another command.
Brings the binary PFT v1 replayer (candle/pft/) onto current master. The branch only adds files under candle/pft/; the ocaml.ml and FFI additions it needed are already on master. Bot: opus (Claude Opus 5.5) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
sequent-bot
Bot
force-pushed
the
bot/opus/pft-replay
branch
from
October 27, 2026 10:27
bfa064b to
379758f
Compare
xrchz
marked this pull request as ready for review
October 27, 2026 10:52
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Proposes merging the
pftbranch, the PFT v1 proof-trace replayer, intomaster, so that replaying HOL4-produced proof traces in Candle lives on the main line rather than on a long-lived branch.This is a plain merge of
pftinto currentmaster, with no conflicts and no changes to thepftwork itself: the files undercandle/pft/are byte-identical topft's. Againstmasterthe net change is four new files:candle/pft/replay.ml: the replayer (reads a binary PFT v1 trace with thecandleruleset and replays it through Candle's kernel, including compute commands;replay_allreplays several files in order, keeping saved theorems between them)candle/pft/pft.ml: loads the kernel and early HOL Light files needed for replay, in place ofhol.mlcandle/pft/misc.ml,candle/pft/extract_footer: helpersNo HOL Light file,
candle/kernel.ml,ocaml.mlor the FFI patch changes: theocaml.mlandbasis_ffi.c.patchadditionspftmade were already merged intomasterwith #10. Everything new stays insidecandle/pft/, so syncs with upstream HOL Light are unaffected. History is a merge, so thepftcommits keep their authorship.As on
pft, loadingreplay.mlends by replayingcandle/pft/merged.candle.pft.bin(paths are relative to the repository root).Checking. No local build; the repository's regression workflow (
./test.sh) checks that the merged tree still builds and passes the existing suite. The replayer itself isn't exercised by that workflow.🤖 Generated with Claude Code