Skip to content

[doc] AGENTS.md: add XQSuite-vs-Java test-selection rubric; remove stale Known Issues #6483

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
line-o merged 1 commit into from
Jun 15, 2026
Merged

[doc] AGENTS.md: add XQSuite-vs-Java test-selection rubric; remove stale Known Issues #6483

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
23 changes: 17 additions & 6 deletions AGENTS.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -159,6 +159,23 @@ Add to `ErrorCodes.java` in a labeled block for your feature area:
public static final ErrorCode FOXX0001 = new ErrorCode("FOXX0001", "Description");
```

### Choosing XQSuite vs Java tests

Default to **XQSuite** (`%test:` annotations) for anything that is XQuery-level behavior — it's idiomatic, runs in-process, and lives beside the XQuery code.

Use **Java** only when XQSuite structurally can't express or exercise the behavior:

1. The unit under test is **Java, not XQuery** (a util/algorithm class) — pure JUnit.
2. The function needs a **context XQSuite doesn't provide** — above all an HTTP request/response context. `request:`/`response:`/`session:` functions throw `XPDY0002` with no live request. Test via Java with a mocked `RequestWrapper` + `context.setHttpContext(...)` (see `GetData2Test`), or over real HTTP (`RESTServiceTest`).
3. The behavior **IS the HTTP/transport layer** — status codes, response headers (e.g. `Content-Type`), serialization wire format, end-to-end content negotiation → Java HTTP integration test (`RESTServiceTest`).
4. Behavior depends on **Java-level wiring** — broker pool, locking/concurrency, transactions, startup/config.

Within Java, use the lightest vehicle that exercises the real behavior: pure unit test for pure logic; mocked-request unit test for request-bound function logic (`GetData2Test` pattern); full HTTP integration test only when you need the real request pipeline / transport.

One-line test: "Can this be a pure XQuery assertion, runnable without an HTTP request or Java-internal state?" → XQSuite. Otherwise → Java, lightest form.

Concrete precedent: PR eXist-db/exist#6477 (request-module content negotiation) — `request:negotiate-content-type` / `request:parse-accept-header` couldn't be XQSuite-tested (request-bound), so they use `AcceptHeaderTest` (pure logic) + `RESTServiceTest` (HTTP wiring).

## Git & PR Workflow

### Remotes
Expand Down Expand Up @@ -216,12 +233,6 @@ eXist-db uses the [exist-xqts-runner](https://github.com/eXist-db/exist-xqts-run
- **BaseX**: reference implementation for XQuery 4.0 features including XQUF and ixml
- **Saxon**: reference implementation for XQuery 4.0, XPath 4.0, and XSLT 4.0

## Known Issues

- `groupby.collation` test is flaky — occasionally throws `ArrayIndexOutOfBoundsException`. Pre-existing FLWOR bug.
- eXist doesn't check HOF return types, only arity (issue #3382). `fn:filter` uses `effectiveBooleanValue()` instead of validating `xs:boolean` return type.
- `fn:doc()` can only load documents from the XML database, not from the local filesystem via `file://` URIs (by design — security boundary).

---

<!-- prethink-context -->
Expand Down