Conversation
`d64607130` ("[GOLD-RING] parser: the Zig subset lexer.t27 needed", a change to `bootstrap/src/compiler.rs`) removed three lines from `cli/tri/src/main.rs` in one hunk: the `mod` declaration, the `Elab` variant and the dispatch arm. The 319-line file stayed. Since then: $ tri elab classify error: unrecognized subcommand 'elab' Nothing said so. `cargo build` cannot fail on a file it does not compile, and the test suite went 358 -> 354 without a gate that reads the number. Four tests left the suite in silence. `tools/check_elab_ratchet.py` still instructs its reader to run this command, in its docstring and in its failure output, and two NOW documents describe it as working. `tri elab` was written because that same script once counted iverilog's per-file summary line `N error(s) during elaboration.` as a diagnostic -- 25 phantoms in a reported 186, quoted onward before anyone read the rows. Restored, it reads: real diagnostics: 176 summary lines NOT counted: 28 Nothing else changes: the file is untouched, and its four tests pass as written. Closes 2900
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
The general form of the defect in the same commit. Rust compiles a file only if
a `mod` declaration leads to it from the crate root; a file nothing declares is
not unused, it is a file the compiler never opens. It cannot warn, cannot fail
to build, and its `#[test]` functions do not exist for `cargo test`.
bootstrap 97 files under src/, 90 compiled, 7 orphaned
bootstrap/src/_enrich/mod.rs 441 lines
bootstrap/src/math_pslq.rs 589 lines
bootstrap/src/neural/attention.rs 4 lines
bootstrap/src/neural/hslm.rs 4 lines
bootstrap/src/runtime_minimal.rs 19 lines
bootstrap/src/runtime_minimal_test.rs 18 lines
bootstrap/src/tooling.rs 409 lines
cli/tri 33 files under src/, 33 compiled, 0 orphaned
Historical control: the same binary run against origin/master, where `mod elab;`
is still missing, prints
cli/tri 32 files under src/, 31 compiled, 1 orphaned
cli/tri/src/elab.rs 319 lines 4 #[test] that cargo cannot see
It fires on the defect and is silent after the fix in the parent commit.
Resolution follows the language rather than a heuristic. `mod c;` in `a/b.rs`
means `a/b/c.rs` or `a/b/c/mod.rs`, never `a/c.rs`; `#[path]` and `include!` are
edges; inline `mod c { }` needs no file. The loose rule -- is this stem named by
any `mod` anywhere -- hides a real orphan behind a same-named module in an
unrelated directory, and its mutation fails the first test here.
`reached` counts files outside `src/` too, because `#[path = "../../gen/..."]`
is a real edge, so the report does not subtract the two and call the difference
orphans; it prints the two comparable sets and says where the rest went.
Refs #2900
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
PR DashboardGenerated at: 2026-08-29 22:22:29 UTC
Summary
Seal Status
|
Contributor
PR DashboardGenerated at: 2026-08-29 22:26:15 UTC
Summary
Seal Status
|
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
PR DashboardGenerated at: 2026-08-29 22:33:02 UTC
Summary
Seal Status
|
The gate that would have caught #2427. $ tri mods orphan --gate # at origin/master ::error::cli/tri: orphaned rose 0 -> 1. A file left the build and nothing else would have said so. exit 1 $ tri mods orphan --gate # after the fix in this branch ORPHAN CEILING: CLEAN exit 0 Exact match in both directions, following MAX_SORRY: a count above its ceiling is a regression, and a count below it is slack the next orphan can hide in. Both were exercised -- a planted file in cli/tri/src exits 1, and a ceiling raised to 3 against an actual 0 exits 1 with the instruction to lower it. `--self-check` is the negative control. It builds a two-file crate in a temp directory, one declared and one not, and exits non-zero unless exactly the undeclared file comes back -- so a green run from this gate means the walk was working, not that it was skipped. The workflow carries NO `paths:` filter, and the comment says why: the change that strands a file is a change to the crate ROOT, not to the file that goes missing. A filter on the stranded path is silent for exactly this defect. bootstrap's 7 orphans are recorded, not repaired. `tooling.rs` alone holds 7 gate bodies (#2905), and connecting any of them is a decision about what goes red rather than a repair. The ledger says so in the file, beside the number. Refs #2900 Refs #2905
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
1 similar comment
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
PR DashboardGenerated at: 2026-08-29 22:33:43 UTC
Summary
Seal Status
|
This was referenced Aug 29, 2026
Owner
Author
|
Superseded by a clean-history branch: the first commit here said |
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.
tri elabreportsunrecognized subcommand. Its 319-line implementation is inthe tree; nothing declares it.
d64607130(#2427, a Zig-lexer change tobootstrap/src/compiler.rs) removedthree lines from
main.rsin a single hunk —mod elab;, theElabvariant,the dispatch arm — and left the file. This restores exactly those three lines.
What nothing printed
cargo buildcargo testsrc/Four tests left the suite without a line of output. The count went down
silently, which is #2895 in another language: nothing here prints what a build
skipped.
Why this command in particular
tri elabexists becausetools/check_elab_ratchet.pycounted every stderr linecontaining
" error", and iverilog closes each failing file withN error(s) during elaboration.— a total, which matched. 25 phantoms in a reported 186,and the number reached commit messages, an issue comment and a status page.
That script still points at the command it lost:
After
elab.rsis not modified. Its four tests pass as written.Closes #2900