feat(tri): local wave commands that work without a built compiler - #2244
Merged
Conversation
gHashTag
enabled auto-merge (squash)
August 19, 2026 16:48
This was referenced Aug 19, 2026
gHashTag
pushed a commit
that referenced
this pull request
Aug 19, 2026
… false alarms are lesson 1402 (Refs #1959) tri wave/disk/ci/lesson/theorem dispatch before t27c is resolved, so they work on a machine where the compiler is not built. Landed to master as #2244. The numbering checkers reported 180+ collisions on their first run; every one was their own regex (T709 matched inside T709a, lesson 88 inside **88.03%**). Lesson 1402 records it -- written by tri lesson itself. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This was referenced Aug 19, 2026
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
This was referenced Aug 19, 2026
…oses #2243) tri resolved t27c before dispatching anything, so every subcommand died with "t27c not found" on a machine where the compiler is not built -- the exact state a machine is in when the compiler is what you are fixing. wave/disk/ci/lesson/theorem now dispatch first and need no compiler. disk is a free-space gate (ENOSPC killed two autonomous sessions). lesson/theorem --check audit numbering; the first version reported 180+ false collisions because it matched T709 inside T709a and lesson 88 inside **88.03%** -- fixed to anchor on the whole identifier, and both corpora then come back clean. Passthrough unchanged. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
An unattended loop checks the same things every cycle: free space above the
floor, tree clean, work pushed, lesson and theorem numbering intact. `tri audit`
runs all of them and exits non-zero on a real failure, so a cron wave can gate
on it instead of re-deriving the checks by hand.
The push check compares against origin/<branch>, not @{upstream}: a worktree
branch is frequently configured to track master, and its first version reported
"62 commits not pushed" -- a true count against the wrong ref. It now prints the
ref it compared against, because a count without its reference point is noise.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
'Have I learned this before' is the question an unattended loop asks most often and answers worst. tri recall greps the HEADINGS only, so the answer fits on a screen instead of dumping the corpus. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The most-used operation of the last three waves of area work, and it never needed the compiler: LUT, CARRY4, MUXF and FF counts for a top, from local yosys. Counting CARRY4 alongside LUT is the point -- a constant-add decoder lands entirely on the carry chain and a LUT-only census reports it as free. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The loop re-reads its own conclusions constantly and the two files are 12k and 28k lines, so tail-with-context was the wrong tool. tri last prints the headings only. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
… reports Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
gHashTag
force-pushed
the
tri-wave-commands-w935
branch
from
August 20, 2026 00:36
d7ba747 to
0b1eb3b
Compare
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
PR DashboardGenerated at: 2026-08-20 00:36:55 UTC
Summary
Seal Status
|
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
2 participants
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.
Closes #2243
triresolved t27c before dispatching anything, so every subcommand died with "t27c not found" on a machine where the compiler is not built -- the exact state a machine is in when the compiler is what you are fixing. Five local commands now dispatch first:tri wavetri disk [--min GIB]tri ci [SLUG ...]tri lesson "T" "B",--checktri theorem [--check]The checkers found their own bug first.
theorem --checkv1 reported 180+ duplicate theorems andlesson --checktwo out-of-order lessons. All false:T709a/T709bare distinct sub-theorems (the regex matchedT709inside them), and**88.03% on Fashion**is bold prose, not lesson 88. Anchored on the whole identifier --T[0-9]+[a-z]*, and the space only a real header has -- both corpora come back clean: 997 theorem headings, monotonic, zero collisions; 629 lessons, no gaps.Verified:
bash -n; every subcommand exercised;tri disk --min 999exits 1;tri cilisted live runs in both repos;tri lessonwrote its own lesson about the false collisions. Compiler passthrough byte-identical (tri test->t27c suite --repo-root).🤖 Generated with Claude Code