Skip to content

docs: changelog and version scoping for 1.0.1 - #158

Merged
luisleo526 merged 2 commits into
mainfrom
rel/1.0.1-notes
Oct 2, 2026
Merged

luisleo526 merged 2 commits into
mainfrom
rel/1.0.1-notes

Conversation

@luisleo526

Copy link
Copy Markdown
Contributor

Docs only, for the 1.0.1 pair release: codegen 1.0.1 with engine v1.0.1. No code, test, workflow, package metadata or VERSION change; the release workflow writes VERSION.

What changed

Checked

Run on 67892cd (this branch's code) and on v1.0.0, CPython 3.14.6, Apple M4 Max:

  • The engine corpus's 325 sources (as v1.0.0 pins it) and the 277 gate fixtures transpile to byte-identical C++; the 13 refused fixtures are refused with the same message.
  • The largest transpilation takes 0.048 s (0.049 s on v1.0.0); the largest source is 9,869 characters. The Limits table stands.
  • Probes of the refusals: a[1] + 1 and c.push(a[1]) on an array transpile on 1.0.0 (with the warning "array history indexing uses the current collection element") and are refused on 1.0.1 (CE10123); c[1].v (CE10011), c.v[1] (CE10290) and m[1].get(0, 0) are refused; (c[1]).v, (a[1]).size(), (m[1]).get(0, 0) and Cell.new(v = bar_index) transpile; (close * 2).m() still emits None.
  • The @pf-trace example gives the same C++ and the /* unsupported: ta.rsi */ marker on both.
  • The quick-start SMA cross transpiles to the same C++ on both. Run with engine 6b45f510 (v1.0.0 plus two docs commits; v1.0.1's code) through the engine tutorial (Linux arm64, GCC 13), it prints the README's block: 9 trades (6W / 3L), +738.20, +679.22 / -335.02, -788.79. With initial_capital=1000000, default_qty_type=strategy.fixed, default_qty_value=1 it books 13 trades.
  • python -m build and twine check --strict pass.
  • The 1,384-source byte identity is the two PRs' own measurement (History of user-defined objects and drawing references #156: identical to 01b4ee3; History of arrays and matrices; int into a float field #157: identical to its base in fresh processes).

No sweep: docs only.

Notes for review

  • The changelog policy says to leave a planned release Unreleased until its tag exists. This section carries the planned date, 2026-10-02, because the release runs right after this merges and the sdist ships this CHANGELOG.
  • The engine src/source/ link and the clone command name v1.0.1, which exists once the engine release has run; the engine releases first.

🤖 Generated with Claude Code

luisleo526 and others added 2 commits October 2, 2026 10:10
1.0.1 pairs with engine v1.0.1, which changes only documentation since
v1.0.0. The changelog section covers #156 and #157, the changes since
1.0.0 that a user can observe:

- C++ that did not compile now compiles: the history of user-defined
  objects and drawings (#156) and of arrays and matrices (#157), and a
  non-constant int given to a float field (#157);
- one value that came out silently wrong: an int na assigned to a float
  field was stored as -2147483648 and is na now (#157);
- new compile-time refusals with TradingView's codes (CE10010, CE10011,
  CE10101, CE10122, CE10123, CE10173, CE10187, CE10290, CE10009), and the
  forms refused by name;
- the "not covered" limits of both pull requests, the migration step
  (regenerate C++ and relink) and the pairing (1.0.1 supports only engine
  v1.0.1). The Python and JSON contract and the report keys do not change.

The README, CONTRIBUTING, AGENTS.md / CLAUDE.md, docs/PUBLIC_CONTRACT.md
and docs/pine-cap-activation.md name 1.0.1 where they named 1.0.0 as the
current release or pair, as the README's release-lane note asks (the
release section, the baseline paragraph, the engine src/source link, the
@pf-trace note, the clone command and its output, the timing note, the
explicit-attachment section, a 1.0.1 row in the pairing table). Lines
that say what changed in 1.0.0 stay, and no line calls a version the
latest.

Checked on this tree's code (67892cd) against v1.0.0, CPython 3.14.6 on
an Apple M4 Max: the engine corpus's 325 sources and the 277 gate fixtures
transpile to byte-identical C++ (13 refusals, identical messages); the
largest transpilation takes 0.048 s (0.049 s on v1.0.0); the quick-start
SMA cross transpiles to the same C++. The tutorial run with that C++
against engine 6b45f510 (Linux arm64, GCC 13) prints the README's 9 trades,
+738.20, and 13 trades with the old sizing declared.

Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
@luisleo526
luisleo526 merged commit fd8e5f2 into main Oct 2, 2026
9 checks passed
@luisleo526
luisleo526 deleted the rel/1.0.1-notes branch October 2, 2026 03:08
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