Accepted is not the same as accepted on the whole spec

[gHashTag]

Refs #2754

The corpus table says "217 specs produce Zig that Zig takes". None of its columns has ever said on how much of the spec — and 87 specs are accepted while part of their text is thrown away. The thrown-away part is the bodies of invariants, which is to say the assertions.

  ALL FOUR accept               63    9.7%

  specs with tokens DROPPED     87   13.4%
    ... tokens dropped       32485
  Accepted is not the same as accepted ON THE WHOLE SPEC.

--per-spec gains a dropped column, so the diff that names a moved row also shows what that row throws away.

The competitive read: docs/PARSER_DISCARD_LANDSCAPE.md

Three designs that do not have this failure mode, and what each costs:

	how the dropped text is handled	cost
rowan — rust-analyzer, Swift, C#, Kotlin	discarding is unrepresentable: the source must be perfectly reconstructible from the tree, errors and all	two tree layers; every consumer must tolerate error nodes rather than assume a well-formed AST
tree-sitter	kept as an ERROR node, addressable by query; zero-width MISSING nodes where insertion is the cheaper repair	recovery costs are opaque by its own documentation — and tree-sitter parse still does not report an error when an ERROR node is present (#4049)
bison error productions	the grammar names its own recovery points	designed in per-construct; silent about how much was skipped at runtime

The middle row is the one worth reading twice. Tree-sitter has the representation and still does not report it — the same defect this repository keeps finding from the other direction, in a project with far more users.

t27 sits closer to tree-sitter than to rowan: recovery happens and the discarded region is not in the AST. It does one thing more than tree-sitter's CLI — it counts the discard (parse-complete) and ratchets it (parse-no-discard is a suite phase). The note is explicit that it is not a proposal to rewrite the parser as a lossless CST, and names the cheap half of rowan's property that was available and unused. This PR writes the third item on that list, so the document does not describe an intention.

Skill sections 150–153

• when a column moves, add the command that names the rows — one flag beat a hand-rolled harness that had already lied once
• two node shapes reaching one emitter arm, and only one of them was read
• prove a wrapper is dead before removing it: 0 of 156 C-accepted specs contained .v = {
• a sync command must recompute the truth, not pick the newer file — the rewrite is what found the 31 dangling pairs

🤖 Generated with Claude Code

[github-actions]

PR Dashboard

Generated at: 2026-08-29 00:03:06 UTC

• Accepted is not the same as accepted on the whole spec #2770 Accepted is not the same as accepted on the whole spec | Created: 2026-08-29 | Updated: 2026-08-29 | 13 failing / 25 total
• Land the formal verification harness on master (it has never run) #2727 Land the formal verification harness on master (it has never run) | Created: 2026-08-27 | Updated: 2026-08-27 | 5 failing / 24 total
• chore(deps)(deps): bump thiserror from 2.0.18 to 2.0.20 #2693 chore(deps)(deps): bump thiserror from 2.0.18 to 2.0.20 | Created: 2026-08-24 | Updated: 2026-08-28 | 3 failing / 34 total
• chore(deps)(deps): bump hyper from 1.10.1 to 1.11.0 #2692 chore(deps)(deps): bump hyper from 1.10.1 to 1.11.0 | Created: 2026-08-24 | Updated: 2026-08-28 | 3 failing / 33 total
• chore(deps)(deps): bump base64 from 0.22.1 to 0.23.1 #2691 chore(deps)(deps): bump base64 from 0.22.1 to 0.23.1 | Created: 2026-08-24 | Updated: 2026-08-28 | 3 failing / 34 total
• chore(deps)(deps): bump serde from 1.0.228 to 1.0.229 #2690 chore(deps)(deps): bump serde from 1.0.228 to 1.0.229 | Created: 2026-08-24 | Updated: 2026-08-28 | 3 failing / 34 total
• chore(deps)(deps): bump uuid from 1.23.2 to 1.25.0 #2689 chore(deps)(deps): bump uuid from 1.23.2 to 1.25.0 | Created: 2026-08-24 | Updated: 2026-08-28 | 3 failing / 34 total
• Unbreak FPGA CI and correct two unbacked hardware claims #2081 Unbreak FPGA CI and correct two unbacked hardware claims | Created: 2026-08-11 | Updated: 2026-08-20 | 5 failing / 27 total

Summary

Status	Count
Total Open PRs	8
PRs with Failing Checks	7
PRs with All Checks Green	1
READY	0
FAILING	7
PENDING	0

Seal Status

• ⚠️ STALE -- sha256(compiler.rs)=3abc96dcebd6 != manifest seal=87e5cbd3ad94.
  The committed NMSE numbers were certified against an older compiler.rs.
  Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

[github-actions]

📓 NotebookLM Notebook linked to this PR

• Notebook ID: b83263109fb055dc
• URL: https://notebooklm.google.com/notebook/b83263109fb055dc

This notebook contains session context, decisions, and artifacts for this work.