# `testing/conformance` — Apollo-bound conformance harness

<!-- begin:auto-docs -->
_(no submodules)_
<!-- end:auto-docs -->

## Purpose

`testing/conformance` is the re-export of `runGraphqlEngineConformance`
from `@suaveplan/api-graphql/testing`, pre-bound to the Apollo factory.
Consumers call `runApolloConformance()` and the harness exercises the
abstract `GraphqlEngine` contract against the concrete Apollo adapter.
Both `api-graphql-apollo` and `api-graphql-yoga` MUST pass the same
invariants — operator behaviour is engine-agnostic at the contract
boundary.

The module also re-exports the unmodified `runGraphqlEngineConformance`
so downstream packages that want to run the harness against a
*different* factory (e.g. a federated subgraph composing Apollo + a
custom transport) keep a single import surface.

## Features

- One-line invocation — `runApolloConformance()` runs every scenario.
- Pre-bound factory — callers do not need to know Apollo's
  construction args.
- Labelled — the second-positional `label` parameter shows up in
  test output so split CI logs are unambiguous.
- Re-exports `runGraphqlEngineConformance` for advanced callers.

## Usage: Basic

```ts
import { runApolloConformance } from "@suaveplan/api-graphql-apollo/testing";

runApolloConformance();
```

## Usage: Advanced

Run the conformance suite under a custom label so two adapter variants
appear distinctly in CI output:

```ts
import { runApolloConformance } from "@suaveplan/api-graphql-apollo/testing";

runApolloConformance("apollo (persisted-queries)");
runApolloConformance("apollo (no-persisted-queries)");
```

## API Reference

- `runApolloConformance(label?: string): void`.
- `runGraphqlEngineConformance(opts): void` — unchanged re-export.

## Implementation Notes

The Apollo-bound factory yields a fresh engine per scenario so each
invariant exercises a clean lifecycle. The `makeRequestContext` slot
is the cast `{} as RequestContext` — sufficient for the abstract
contract because the harness never touches context fields beyond
hand-off. Tests that need a richer context construct one explicitly.