Skip to content

experiments: make 'the public API is near-frozen' an explicit rule - #224

Merged
danReynolds merged 1 commit into
mainfrom
docs/public-api-policy
Jul 1, 2026
Merged

experiments: make 'the public API is near-frozen' an explicit rule#224
danReynolds merged 1 commit into
mainfrom
docs/public-api-policy

Conversation

@danReynolds

Copy link
Copy Markdown
Owner

The runner Goal already said "preserving the lean public API," but nothing enforced it — three open experiments grew the public surface (#222 WriteStatement, #223 ReadStatement, #155 IVM types). This states the rule outright.

Rule: delivering a win under the existing public surface is the default. Adding or changing public API (a new exported type, a new Database method, a changed return type) is discouraged and only worth proposing for a massive, broadly-applicable win that genuinely cannot be delivered under the current surface — not a niche or trade-off-shaped win the caller opts into by data shape, and this applies to moonshots too. An experiment that only pays off as new public API and doesn't clear the bar is rejected on those grounds.

  • experiments/RUNNER_INSTRUCTIONS.md — full policy under Goal (### The public API is near-frozen).
  • .claude/skills/resqlite-experiment/SKILL.md — concise mirror + pointer in the contract.

Companion action: closing #222 and #223 under this rule. #155 (IVM) is held for separate evaluation against the "massive win" bar.

🤖 Generated with Claude Code

The runner goal already said "preserving the lean public API," but it wasn't
enforced — three open experiments (208/209 batch APIs, 160 IVM) grew the public
surface. State the rule outright: delivering a win under the existing public
surface is the default; adding or changing public API is discouraged and only
worth proposing for a massive, broadly-applicable win that genuinely can't be
delivered under the current surface (not a niche or trade-off-shaped win, and
this applies to moonshots too). An experiment that only pays off as new public
API and doesn't clear the bar is rejected on those grounds.

Added the full policy to experiments/RUNNER_INSTRUCTIONS.md (Goal) and a concise
mirror + pointer in the resqlite-experiment SKILL.md contract.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings July 1, 2026 20:25

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Pull request overview

This PR makes “the public API is near-frozen” an explicit, written rule for resqlite performance experiments, reinforcing that experiments should default to delivering wins under the existing exported surface and only propose new/changed public API for truly massive, broadly-applicable wins.

Changes:

  • Adds a detailed policy section under the runner Goal in experiments/RUNNER_INSTRUCTIONS.md.
  • Adds a concise mirrored policy section in the resqlite-experiment skill contract with a deep link to the full policy.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File Description
experiments/RUNNER_INSTRUCTIONS.md Documents the near-frozen public API rule as part of the experiment runner’s stated goals and acceptance bar.
.claude/skills/resqlite-experiment/SKILL.md Mirrors the policy in the skill guidance and links to the canonical runner instructions section.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

@danReynolds
danReynolds merged commit 4ade1a2 into main Jul 1, 2026
7 checks passed
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.

2 participants