Skip to content

venue-test: version fixture files and reject empty vector sets#437

Open
mfw78 wants to merge 3 commits into
feat/guard-deny-quotafrom
feat/fold-tail-hygiene
Open

venue-test: version fixture files and reject empty vector sets#437
mfw78 wants to merge 3 commits into
feat/guard-deny-quotafrom
feat/fold-tail-hygiene

Conversation

@mfw78

@mfw78 mfw78 commented Jul 16, 2026

Copy link
Copy Markdown
Contributor

What

Add a FormatVersion discriminator to codec-vector and header-golden fixture files (parse fails closed on an unknown version); reject empty vector/golden sets on parse and on the conformance check.

Why

Cross-language codec/header goldens conform vacuously when empty and carried no tag to catch a stale reader against a future file layout. The migration-cruft deletion and the denied() MUST-NOT-retry doc from the same fold-tail scope are already in tree.

Testing

cargo test -p nexum-venue-test (unit + conformance + doctests, all green).

AI Assistance

Implemented and tested with Claude Code.

Closes #365

mfw78 added 3 commits July 16, 2026 17:51
Fixture files lead with a format version; a reader fails closed on an
unknown tag and refuses an empty vector or golden set, which would
conform vacuously. Published reference fixtures regenerated.

@lgahdl lgahdl left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Verified the two risks that matter here and both check out clean: FormatVersion fails closed identically for codec vectors and header goldens (both route through the same shared fixture::FormatVersion::deserialize, no asymmetry), and the empty-set rejection is genuinely enforced at both parse-time (fixture::non_empty) and check-time (the is_empty() guard in each check()) as two independent gates, not just one dressed up as two. The fixture JSON files got "version": 1" spliced correctly into existing content, and the echo-venue change is test-only. Comments accurately describe the new fail-closed/never-empty invariants. Approving.

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.

packaging: land the fold-tail codec discriminator, migration-cruft deletion and retry doc

2 participants