Skip to content

chore: roadmap hygiene, settle 1.0.0 suite gate (20/29), tostring(fn) fix - #353

Merged
davydog187 merged 3 commits into
mainfrom
chore/roadmap-hygiene
Jun 10, 2026
Merged

davydog187 merged 3 commits into
mainfrom
chore/roadmap-hygiene

Conversation

@davydog187

Copy link
Copy Markdown
Contributor

Why

The project's tracking artifacts had drifted out of sync with reality, and the 1.0.0 suite gate was an unverified guess. This PR re-grounds both in evidence, and lands one correctness fix surfaced along the way.

What

1. Roadmap / plan-status hygiene

ship-a-plan never wrote merged status back to 35 plan files (A19–A48, B8–B14, B5a/b-v2, B9 — all shipped). Flipped each review → merged (every PR confirmed MERGED via gh). Refreshed the stale ROADMAP.md status block (dated 2026-05-21):

  • Unit tests 1,705 → 2,252 passing, 0 failing.
  • Lua 5.3 suite 6/29 → 17/29 passing (the old number predated the A20–A24 triage).
  • Notes rc.2 shipped; current focus is the A13 gates.

Left the In flight / Next / Deferred narrative for human update, per the ROADMAP cadence rule.

2. Settle the 1.0.0 suite gate

A live re-triage of every whole-file skip showed the skip reasons were stale and the speculative ≥22/29 A13 target was never achievable (22 would require coroutines, a full debug library, or a perf rewrite — all deliberately deferred).

Settled: gate = 20/29 passing, 9 documented exclusions.

Category Files
filesystem / subprocess non-goals main, files, attrib, verybig
capability non-goals coroutine, db
perf-bound (>90s), revisit 1.0.x big, closure
PUC error-wording divergence errors

Corrected each skip reason to the real first-failure (e.g. math's math.huge identities actually hold; big/closure confirmed >90s in isolation, reclassified :performance). 17 pass today; path to 20 is three triage candidates (strings/sort/math).

3. Quick win: tostring on a function

tostring(print) returned the bare string "function"; PUC-Lua returns "function: 0x...". Now renders function: 0x<addr> (and function: builtin: 0x<addr> for native functions), mirroring the existing table: 0x... format. type(f) still returns "function". Advances strings.lua past line 126 (next blocker: string.format('%q') at line 153). New Value.to_string/1 unit tests + CHANGELOG entry.

Verification

  • mix compile --warnings-as-errors clean
  • mix test → 2,252 passed, 0 failures, 19 skipped
  • mix format --check-formatted clean
  • Suite stays green: 17 passed, 12 skipped

No production code touched beyond the scoped tostring fix.

The ship-a-plan skill never wrote merged status back to 35 plan files
(A19-A48, B8-B14, B5a/b-v2, B9 all shipped) and the ROADMAP status block
was stale at 2026-05-21. Flip those 35 frontmatter statuses review ->
merged (each PR confirmed MERGED) and refresh the status block:

- Unit tests 1,705 -> 2,230 passing, 0 failing.
- Lua 5.3 suite 6/29 -> 17/29 passing.
- Note rc.2 shipped and current focus is the A13 release gates.

Leaves the In flight / Next / Deferred narrative for human update per
the ROADMAP cadence rule.
A live re-triage of every whole-file-skipped suite file shows the skip
*reasons* were stale and the speculative '>=22/29' A13 target was never
achievable. Settle the gate and correct the manifest to match reality:

- big.lua / closure.lua: confirmed >90s in isolation -> reclassify from
  'pending triage (suspected timeout)' to perf-bound 1.0 exclusions
  (:performance), revisit 1.0.x with B5 / #324.
- coroutine.lua / db.lua: frame as capability non-goals (1.0 exclusion).
- errors.lua: the real blocker is PUC '[string "..."]:N:' parse-error
  wording vs our 'parse error: ...'; documented 1.0 exclusion.
- math.lua: prior 'math.huge stand-in' reason was wrong (huge+1==huge
  and 1/0==huge both hold); real first failure near line 47 unclassified
  -> triage candidate.
- sort.lua: first checkerror passes; a later assertion fails -> triage
  candidate.
- strings.lua: tostring(function) renders 'function' with no ': 0x...'
  address (line 126) -> cheap fix, triage candidate.

A13 suite gate set to 20/29 with 9 documented exclusions (4 fs/subprocess
+ 2 capability + 2 perf + 1 error-wording); 17 pass today, path to 20 is
the strings/sort/math triage plans. Suite stays green (17 passed, 12
skipped). No production code touched.
tostring(print) returned the bare string "function"; PUC-Lua returns
"function: 0x..." (and "function: builtin: ..." for C functions). Lua
code that branches on string.find(tostring(f), 'function:') — including
the official strings.lua suite (line 126) — failed against the bare
form. Render lua/compiled closures as "function: 0x<addr>" and native
functions as "function: builtin: 0x<addr>", mirroring the existing
"table: 0x..." format; the address is a deterministic per-value
pseudo-pointer via :erlang.phash2. type(f) still returns "function".

Advances strings.lua past line 126 (next blocker is string.format('%q')
escaping at line 153); skip reason updated accordingly. No suite count
change yet — strings.lua stays a triage candidate.
@davydog187
davydog187 merged commit 579bee5 into main Jun 10, 2026
5 checks passed
@davydog187
davydog187 deleted the chore/roadmap-hygiene branch June 10, 2026 22:51
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant