Skip to content

fix(zingolib): pass the gates on #3112 and place its changelog entries - #1

Merged
j0ntz merged 34 commits into
EdgeApp:jon/runtime-sapling-paramsfrom
zingolabs:fix/3112-gates
Nov 3, 2026
Merged

j0ntz merged 34 commits into
EdgeApp:jon/runtime-sapling-paramsfrom
zingolabs:fix/3112-gates

Conversation

@Oscar-Pepper

Copy link
Copy Markdown
  • merges dev so the binding manifest check passes
  • moves the changelog entries to unreleased as zingolib v8 has released since
  • adds changelog entries for SaplingParamFile::ALL, file_name and verify
  • fixes cargo fmt checks
  • fixes the test suite to run with the new feature on
  • adds CI suite for testing with the new feature on
  • adds a comment to verify

dependabot Bot and others added 30 commits October 27, 2026 20:58
Bumps [rustls](https://github.com/rustls/rustls) from 0.23.43 to 0.23.46.
- [Release notes](https://github.com/rustls/rustls/releases)
- [Changelog](https://github.com/rustls/rustls/blob/main/CHANGELOG.md)
- [Commits](rustls/rustls@v/0.23.43...v/0.23.46)

---
updated-dependencies:
- dependency-name: rustls
  dependency-version: 0.23.46
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
…37674f7b4d1cc19b87

chore(bindings): record the publication of d496bea
release zingolib v8.0.0 and zingo-cli v0.5.0
…and a forced bump is sized by the dependent's own interface
…s step 6, and zingolib/0064 governs the document
An entry's changelog section follows the newest earlier entry that is an
ancestor of its commit rather than the previous entry, and the publish
plan refuses only a commit none of whose ancestors is in the record. A
release line candidate merged into dev after a later commit of dev was
published is therefore accepted, and its section carries the fix set alone.

Fixes #3113
…tors

The newest earlier entry that is an ancestor of a commit is not the
nearest by ancestry. An older commit of dev published late is an ancestor
of the entries published before it, so every later entry took it as
Since and repeated the lines those entries had already carried. The first
commit of dev after a release line merges back has the line's candidate
and the last entry of dev among its ancestors, neither an ancestor of the
other, and a single Since repeated one side's lines.

recorded_ancestors returns the earlier entries that are ancestors of the
commit and of no other such entry, through git merge-base --independent,
and publication_commits derives each entry's Since commits from it. The
changelog section carries the lines gained against every Since commit and
drops a heading left with nothing beneath it, and its Since line names
every commit. The publish plan accepts a commit when any entry is among
its ancestors, as before. bindings/CHANGELOG.md is regenerated: only the
header differs, so every existing section reproduces byte for byte.

Addresses the two review notes on #3116.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…10e8af12d6d4d29a41

chore(bindings): record the publication of 81d7997
fix(bindings): the publication record accepts a release line candidate
zingo-adrs zingolib/0064 merged on 2026-11-02 and names this procedure as
its specification, so the status line is accepted, the AGENTS.md rule is
dated by the record, and the docs/adr submodule includes it.

zingolib#3116 landed on 2026-10-30, so step 6 states the rule it put in
place instead of saying the procedure does not function until it does.

Addresses the review on #3114.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…thod

A patch line merges into dev once per candidate and never after the
release, since step 7 commits nothing between the last candidate and the
release. The merge-back resolves version and lock conflicts to dev's side,
places the patch section in version order, and adds the fix's lines under
Unreleased only when dev has cut a release since the patched one, because
that release lacks the fix and the next one from dev is the first on its
line to carry it.

No ruleset can forbid a squash or a rebase for one source branch alone, so
step 6 says the merger chooses the merge commit by hand and the pull
request template reminds them.

Addresses points 4 and 5 of the review on #3114.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A template line reminds and a ruleset enforces. Of the last forty pull
requests merged into dev, thirty-nine were merge commits, so constraining
every pull request into dev costs nothing and the release line's tagged
candidate stays reachable from dev by construction. Step 6 and the
template name the ruleset.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
docs(release): draft the procedure for releasing from dev without a stable branch
build(deps): Bump rustls from 0.23.43 to 0.23.46
Oscar-Pepper and others added 4 commits November 3, 2026 10:55
The changelog entries move under [Unreleased]; after the dev merge they sat
inside the cut [8.0.0] section. SaplingParamFile reads its name, length and
digest from one table, as the two match-to-constant methods were an exact
group for the gate. A comment on verify records why it runs before
LocalTxProver::from_bytes.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Under testutils, load() defaults the directory to the build's parameter
cache when none is set. directory_is_set_once names that cache, and the
absent-file case reads a temporary directory, which needs no global.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
cargo-hack compiled the tests gated on the feature and nothing ran them.
test.yaml takes a sapling-params input under which the archive job fetches
the two files with the default build script and uploads them, and each
shard places them in its parameter cache.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@j0ntz
j0ntz merged commit c8ceeb9 into EdgeApp:jon/runtime-sapling-params Nov 3, 2026
@j0ntz j0ntz mentioned this pull request Nov 3, 2026
1 of 2 tasks
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.

3 participants