Repository navigation
egglog-r2il-probe: report the cold egglog phase in µs; coresearch map for D-DRMC - #1705
Conversation
Operator decision 2026-10-22: cold, out-of-fold costs are reported in microseconds and are not gates. e1_cold_cost now prints the egglog phase (setup + saturation + prove-variants): 21 084 / 26 875 / 28 793 µs over three runs, each a median of 5, release, shared 4-vCPU host. Encode (3-4 µs) and lower + execute (6-15 µs) are unchanged. CLAUDE.md records the scoped exception under the "no millisecond results" P0: cold compile and differential-oracle reference arms only, always µs, always labelled not-gated; in-fold paths stay under the rule. The Stage B board entry gains the numbers and drops the open ruling. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01G22yT6htkcdyXsihxxXdrg
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configuration
Warning Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Premise gate split the question into frame identity, finite-population evidence, promotion and rebinding. Five scouts, a 21-row crosswalk and three co-architect passes. Measured on the PR heads: 12/12 and 20/20 tests pass, clippy clean, 17 and 32 rustfmt diffs, new probe workflow red at Format. ADOPT-NOW candidates: the Unknown-before-Counterexamples ordering defect, tautological fixtures, caller-set proof disposition, frame/focus/ scope split, validity as a carrier of R, CI fix. Nothing adopted. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01G22yT6htkcdyXsihxxXdrg
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01G22yT6htkcdyXsihxxXdrg
…tion directions as open Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01G22yT6htkcdyXsihxxXdrg
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01G22yT6htkcdyXsihxxXdrg
Two commits on one branch.
1. Cold costs (operator decision 2026-10-22: report cold costs in microseconds, not gated)
examples/e1_cold_cost.rsnow prints the egglog phase (setup, saturation,prove-variants), which used to be withheld.CLAUDE.mdadds a scoped exception under "no millisecond results". It covers only cold compiles that run outside the fold and differential-oracle reference arms such as MORK inmork_split.rs. These are always reported in µs and always labelled "reported, not gates". In-fold paths stay under the rule.Release build, Xeon @ 2.10 GHz, 4 vCPU, shared host. Three runs, each a median of 5:
2. Board:
/coresearchexploration map for the D-DRMC stack (#1700–#1704).claude/board/entries/2026-10-22-coresearch-drmc-stack.md, plus the regenerated entries index. This is board text only; no code changes. Nothing is adopted by it. Candidates the map grades ADOPT-NOW:universal_on_focuschecks unknown pairs before counterexamples, so a refuted focus is reported asUnknownand its counterexample is hidden.offer_recipetrusts is set by the caller.Checks
cargo fmt --checkandcargo clippy --release --all-targets -D warningspass foregglog-r2il-probe.🤖 Generated with Claude Code
https://claude.ai/code/session_01G22yT6htkcdyXsihxxXdrg