2014-08-22 17:47:18 +02:00
# Test262 Authoring Guidelines
2014-07-21 17:39:10 +02:00
2017-06-13 00:29:58 +02:00
## Source Material
- https://tc39.github.io/ecma262/
- https://github.com/tc39/ecma262/pulls?q=is%3Apr+is%3Aopen+label%3A%22needs+tests%22
+ This is a list of the "needs tests" PRs for changes to the specification.
2014-08-22 17:47:18 +02:00
## Test Case Names
2014-07-21 17:39:10 +02:00
2015-05-25 23:51:55 +02:00
Test cases should be created in files that are named to identify the feature or API that's being tested.
2014-07-21 17:39:10 +02:00
2018-05-02 22:14:31 +02:00
The names should use alphanumeric characters and `.` , `-` , `_` . Otherwise, there is no strict naming convention, but the file names should be human readable, helpful and, ideally, consistent within a single directory. For examples:
2014-07-21 17:39:10 +02:00
2015-02-13 23:39:17 +01:00
- `Math.fround` handling of `Infinity` : `test/built-ins/Math/fround/Math.fround_Infinity.js`
2017-10-06 22:09:39 +02:00
- `Array.prototype.find` use with `Proxy` : `test/built-ins/Array/prototype/find/Array.prototype.find_callable-Proxy-1.js`
2015-02-13 23:39:17 +01:00
- `arguments` implements an `iterator` interface: `test/language/arguments-object/iterator-interface.js`
2014-07-21 17:39:10 +02:00
2017-10-06 22:09:39 +02:00
See the following directory trees for further recommended examples:
- [test/built-ins/Math/fround ](./test/built-ins/Math/fround/ )
- [test/built-ins/WeakMap ](./test/built-ins/WeakMap/ )
- [test/language/arguments-object ](./test/language/arguments-object/ )
2017-12-10 20:24:16 +01:00
**Note** The project is currently transitioning from a naming system based on specification section numbers. There remains a substantial number of tests that conform to this outdated convention; contributors should ignore that approach when introducing new tests and instead encode this information using [the `esid` frontmatter tag ](#esid ).
2014-07-30 00:27:20 +02:00
2014-08-22 17:47:18 +02:00
## Test Case Style
2014-07-30 00:27:20 +02:00
A test file has three sections: Copyright, Frontmatter, and Body. A test looks roughly like this:
2014-07-04 02:13:42 +02:00
```javascript
2015-02-17 17:54:51 +01:00
// Copyright (C) 2015 [Contributor Name]. All rights reserved.
2014-07-04 02:13:42 +02:00
// This code is governed by the BSD license found in the LICENSE file.
2015-05-25 23:51:55 +02:00
/*---
2014-08-22 17:47:18 +02:00
description: brief description
2014-07-30 00:27:20 +02:00
info: >
verbose test description, multiple lines OK.
2017-10-04 23:45:03 +02:00
(info typically contains relevant, direct quote from ECMAScript)
2014-07-30 00:27:20 +02:00
---*/
2015-05-25 23:51:55 +02:00
2014-07-04 02:13:42 +02:00
[Test Code]
```
2014-07-21 17:39:10 +02:00
2014-08-22 17:47:18 +02:00
### Copyright
2014-07-30 00:27:20 +02:00
The copyright block must be the first section of the test. The copyright block must use `//` style comments.
2014-08-22 17:47:18 +02:00
### Frontmatter
2014-07-30 00:27:20 +02:00
2015-05-25 23:51:55 +02:00
The Test262 frontmatter is a string of [YAML ](https://en.wikipedia.org/wiki/YAML ) enclosed by the comment start tag `/*---` and end tag `---*/` . There must be exactly one Frontmatter per test.
2014-07-30 00:27:20 +02:00
2015-05-25 23:51:55 +02:00
Test262 supports the following tags:
2014-07-30 00:27:20 +02:00
2015-05-25 23:51:55 +02:00
- [**description** ](#description ) (required)
2014-08-22 17:47:18 +02:00
- [**info** ](#info )
- [**negative** ](#negative )
2017-06-28 18:45:25 +02:00
- [**esid** ](#esid ) (required for new tests)
2014-08-22 17:47:18 +02:00
- [**includes** ](#includes )
- [**timeout** ](#timeout )
- [**author** ](#author )
- [**flags** ](#flags )
2017-10-17 22:38:39 +02:00
- [**features** ](#features ) (required for new tests written for new features)
2014-08-22 17:47:18 +02:00
2017-10-06 22:09:39 +02:00
The following tags are deprecated, but exist in old tests:
- [**es5id** ](#es5id ) (deprecated)
- [**es6id** ](#es6id ) (deprecated)
2014-08-22 17:47:18 +02:00
#### description
**description**: [string]
2017-10-06 22:09:39 +02:00
This is one of two required frontmatter tags. The description should be a short, one-line description of the purpose of this testcase. We suggested that the description be kept to less than 100 characters, but clarity is preferred over brevity.
2014-07-21 17:39:10 +02:00
2014-07-04 02:13:42 +02:00
Eg: Insert < LS> between chunks of one string
2014-07-21 17:39:10 +02:00
2014-08-22 17:47:18 +02:00
#### info
**info**: [multiline string]
2017-10-04 23:45:03 +02:00
This allows a long, free-form comment. The comment is almost always a direct
quote from ECMAScript. It is used to indicate the observable being tested
within the file.
2014-08-22 17:47:18 +02:00
2017-10-04 23:45:03 +02:00
For example:
/*---
esid: sec-weakset.prototype.has
description: Throws TypeError if `this` is not Object.
info: >
WeakSet.prototype.has ( value )
1. Let S be the this value.
2. If Type(S) is not Object, throw a TypeError exception.
---*/
Note: Adding more context than the direct quote from ECMAScript should rarely
be necessary. If you must add context to the quote, use the JavaScript
single line comment syntax.
2014-08-22 17:47:18 +02:00
#### negative
2016-03-13 18:26:10 +01:00
**negative**: [dictionary containing **phase** and **type** ]
2014-07-21 17:39:10 +02:00
2014-07-31 22:41:08 +02:00
This means the test is expected to throw an error of the given type. If no error is thrown, a test failure is reported.
2014-07-21 17:39:10 +02:00
2016-03-13 18:26:10 +01:00
- **type**- If an error is thrown, it is implicitly converted to a string. In order for the test to pass, this value must match the name of the error constructor.
2017-12-03 06:20:06 +01:00
- **phase** - Negative tests whose **phase** value is "parse" must produce the specified error prior to executing code. The value "resolution" indicates that the error is expected to result while performing ES2015 module resolution. The value "runtime" dictates that the error is expected to be produced as a result of executing the test code.
2014-07-30 00:27:20 +02:00
2017-10-17 22:45:37 +02:00
For best practices on how to use the negative tag please see [Handling Errors and Negative Test Cases ](#handling-errors-and-negative-test-cases ), below.
2014-07-30 00:27:20 +02:00
2016-03-13 18:26:10 +01:00
For example:
negative:
2017-12-03 06:06:42 +01:00
phase: parse
2016-03-13 18:26:10 +01:00
type: ReferenceError
2016-02-22 15:47:03 +01:00
#### esid
**esid**: [spec-id]
2016-01-28 21:55:50 +01:00
2017-06-27 23:29:22 +02:00
This tag is required for all new feature tests. This tag identifies the hash ID from the portion of the ECMAScript draft which is most recent to the date the test was added. It represents the anchors on the generated HTML version of the specs. E.g.: `esid: sec-typedarray-length` . This tag might be used to replace a `es6id` or further.
2016-01-28 21:55:50 +01:00
2016-01-29 18:29:09 +01:00
When writing a new test for a Stage 3+ spec not yet published on the draft, the `pending` value can be used while a hash ID is not available.
2014-07-30 00:27:20 +02:00
2014-08-22 17:47:18 +02:00
#### includes
**includes**: [file-list]
This tag names a list of helper files that will be included in the test environment prior to running the test. Filenames **must** include the `.js` extension.
2014-07-30 00:27:20 +02:00
2017-12-21 18:17:51 +01:00
The helper files are found in the `harness/` directory. When some code is used repeatedly across a group of tests, a new helper function (or group of helpers) can be defined. Helpers increase test complexity, so they should be created and used sparingly.
2014-07-31 22:41:08 +02:00
2014-08-22 17:47:18 +02:00
#### timeout
**timeout**: [integer]
This tag specifies the number of milliseconds to wait before the test runner declares an [asynchronous test ](#writing-asynchronous-tests ) to have timed out. It has no effect on synchronous tests.
2015-05-25 23:51:55 +02:00
Test authors **should not** use this tag except as a last resort. Each runner is allowed to provide its own default timeout, and the user may be permitted to override this in order to account for unusually fast or slow hardware, network delays, etc.
2014-08-22 17:47:18 +02:00
#### author
**author**: [string]
2015-05-25 23:51:55 +02:00
This tag is used to identify the author of a test case.
2014-08-22 17:47:18 +02:00
#### flags
**flags**: [list]
2014-07-30 00:27:20 +02:00
This tag is for boolean properties associated with the test.
2015-06-08 23:05:22 +02:00
- **`onlyStrict`** - only run the test in strict mode
2015-05-25 23:51:55 +02:00
- **`noStrict`** - only run the test in "sloppy" mode
2015-05-28 20:41:22 +02:00
- **`module`** - interpret the source text as [module
2017-04-14 18:19:39 +02:00
code](https://tc39.github.io/ecma262/#sec-modules)
2015-06-08 21:35:16 +02:00
- **`raw`** - execute the test without any modification (no helpers will be
available); necessary to test the behavior of directive prologue; implies
`noStrict`
2016-02-12 18:59:51 +01:00
- **`async`** - defer interpretation of test results until after the invocation
2016-09-27 23:05:08 +02:00
of the global `$DONE` function
2017-03-06 18:12:42 +01:00
- **`generated`** - informative flag used to denote test files that were
created procedurally using the project's test generation tool; refer to the
section titled "Procedurally-generated tests" for more information on this
process
Make Atomics tests more resilient against intermittent failures (#1617)
Also fixes multiple issues and bugs in various Atomics tests.
CONTRIBUTING.md and INTERPRETING.md:
- Add missing description for `CanBlockIsFalse` flag and also introduce `CanBlockIsTrue` flag which is needed for some tests (see below).
harness/atomicsHelper.js:
- Updated `$262.agent.waitUntil` to clarify it can also be called with BigInt64Array objects.
- Added `$262.agent.timeouts` constants to unify the various, inconsistently used timeouts in the Atomics tests. Each timeout constant has a brief description and a usage example to clarify when it should or should not be used. The default values were tested in SpiderMonkey (locally and CI) under various system load levels and should hopefully also be valid for other engines.
- Added `$262.agent.tryYield` to yield control from the main thread. The default implementation simply calls `$262.agent.sleep`, but test262 hosts could theoretically provide a different implementation.
- Added `$262.agent.trySleep` to replace direct calls to `$262.agent.sleep` from the main thread. Motivation for this function: Some test262 hosts (like browsers) may be able to pause and sleep on the main thread, so they could provide their implementation for `trySleep` which performs a busy-wait or something do nothing.
harness/testAtomics.js:
- Replace `let`, `arrow-functions`, and `for-of` loops to avoid creating extremely long `features` lists in tests using this helper. Removed `Map`, `WeakMap`, `Set`, and `WeakSet` (these weren't even listed in features.yaml) for the same reason and added the missing `Symbol` entry for this file to features.yaml.
- Updated all files including "harness/testAtomics.js" to match the new feature requirements.
test/built-ins/Atomics/store/good-views.js and test/built-ins/Atomics/store/bigint/good-views.js:
- Replace `arrow-functions` and `for-of` loops with ES5 alternatives.
test/built-ins/Atomics/wait/bigint/*.js and test/built-ins/Atomics/wake/bigint/*.js
- Some tests were using `BigInt64Array.BYTES_PER_ELEMENT * 8` for the SharedArrayBuffer length, but their non-BigInt counterparts are using `Int32Array.BYTES_PER_ELEMENT * 4`. For consistency and to make it easier to compare the BigInt against the non-BigInt versions, I've changed it to `BigInt64Array.BYTES_PER_ELEMENT * 4`.
- Also aligned formatting and statement placement when they differed between the non-BigInt and the BigInt version of a test file. (I've diffed some of the non-BigInt and BigInt files against each other and different formatting was a nuisance.)
Test files using `$262.agent.monotonicNow()`:
- Moved `$262.agent.report()` calls outside of the block of code measured `$262.agent.monotonicNow()` to avoid measuring how long it takes to execute `$262.agent.report()`.
- Without this change some tests failed intermittently in certain test configurations in SpiderMonkey. For example with the flags `--ion-eager -- ion-offthread-compile=off` which forces early Ion compilation on the main thread. The `$262.agent.report()` implementation in the SpiderMonkey test262 host embedding uses a for-loop which was forcefully Ion compiled under these settings. And because Ion compilation can take some time, the test case ran longer than `$262.agent.MAX_TIME_EPSILON` which lead to intermittent failures.
Test files using `CanBlockIsFalse` / `CanBlockIsTrue`:
- Some of these tests actually expected that the main thread can wait and [[CanBlock]] is `true` for the agent record executing the test. Therefore I've added a new `CanBlockIsTrue` flag and replaced the flags where needed.
test/built-ins/Atomics/wait/**/*.js and test/built-ins/Atomics/wake/**/*.js:
- Use an atomic counter `RUNNING` in more tests to have better control when a worker agent was actually started.
- Replace the various `$262.agent.sleep(/* Sleep X ms to ensure worker actually sleeps */)` calls with the new `$262.agent.tryYield()` function. This `X` was sometimes as low as 10 milliseconds, which is definitely too short for CI systems under heavy load (observed by intermittent CI failures for SpiderMonkey) and sometimes as high as 500 milliseconds, which is probably much longer than needed even when the system is under heavy load.
- Removed duplicate strings in assertion messages, presumably from copy-pasting the messages between different files.
- Removed extra empty lines at the end of multiple files.
test/built-ins/Atomics/wake/bad-range.js, test/built-ins/Atomics/wake/bigint/bad-range.js, and test/built-ins/Atomics/wait/bigint/non-bigint64-typedarray-throws.js:
- Removed unnecessary `features` and `includes` from this file.
test/built-ins/Atomics/wait/waiterlist-order-of-operations-is-fifo.js and test/built-ins/Atomics/wait/bigint/waiterlist-order-of-operations-is-fifo.js:
- The test was actually broken and didn't test what it said it does. This probably explains #1530.
- The test wants to ensure the waiterlist is implemented as a FIFO structure. This requires that the waiting agents all wait on the same index position, because the waiterlists are defined by each index. But if the agents wait on different indices, each agent is inserted into a different FIFO structure and therefore we can't observe any FIFO ordering between the agents when they're woken up.
- All this requires a bit of synchronization between the main agent and the waiting agent, I hope the added comments help to review these changes.
test/built-ins/Atomics/wait/good-views.js:
- The agent sends multiple reports to the main agent, but only the first one was read.
- This error was introduced during previous refactorings. I've changed it to back to use a while-loop as used in the first version of this file.
test/built-ins/Atomics/wait/no-spurious-wakeup-no-operation.js:
- Only the BigInt version of this test was present, copied it so we also get code coverage for the non-BigInt case.
test/built-ins/Atomics/wait/waiterlist-block-indexedposition-wake.js:
- Added extra while loops to avoid intermittent failures when the agent worker haven't started to wait.
- This should help to avoid some of the intermittent failures we saw for SpiderMonkey.
test/built-ins/Atomics/wake/wake-all-on-loc.js and test/built-ins/Atomics/wake/bigint/wake-all-on-loc.js:
- This test was also no longer after previous refactoring sessions.
- The "B" agent only waited for 10 milliseconds, which made it likely that it already timed out before the main agent was able to call `Atomics.wake`, which in turn rendered the test useless, because the test case wants to ensure that `Atomics.wake` cannot wake "B". But if "B" was already timed out, it can trivially not be woken by `Atomics.wake`.
- Added some safety measure to catch the case when "B" timed out before `Atomics.wake` was called and made it a test error if that happens.
test/built-ins/Atomics/wake/count-defaults-to-infinity-missing.js, test/built-ins/Atomics/wake/count-defaults-to-infinity-undefined.js, and test/built-ins/Atomics/wake/undefined-index-defaults-to-zero.js:
- Changed the `$262.agent.start()` calls to use a for-loop to avoid code duplication.
- (Forgot to undo the code formatting around `assert.sameValue`, still need to change it back.)
- Also more while-loops around `Atomics.wake`.
- These changes should fix #1529 and #1566.
test/built-ins/Atomics/wake/wake-all.js:
- Removed "B" worker agent.
- Without this removal the test case would be exactly equal to test/built-ins/Atomics/wake/wake-all-on-loc.js.
test/built-ins/Atomics/wake/wake-in-order-one-time.js:
- Add for-loops to avoid code duplication and make the test more readable.
- Make the `Atomics.wake` assertion messages unique by adding the current loop counter.
- Add `$262.agent.tryYield()` to give the worker agents enough time to actually start waiting.
test/built-ins/Atomics/wake/wake-in-order.js:
- Removed the outer loop `attempt < 10` because it uses `$262.agent` in a way currently not required to work. And which actually also doesn't work in SpiderMonkey's implementation of `$262.agent`.
- According to INTERPRETING.md `$262.agent.broadcast()` broadcasts its message to all agents ever started by `$262.agent.start()` and then blocks until all agents received the broadcast. It is not required that the agents started by the first `$262.agent.broadcast()` call will all be disabled/destroyed/whatever when the second `$262.agent.broadcast()` call occurs, which then means the second `$262.agent.broadcast()` call still tries to reach the agents started in the first loop iteration, but these may no longer accept broadcasts and therefore won't acknowledge they've received the broadcast. Which then means the second `$262.agent.broadcast()` call waits forever.
2018-07-03 18:34:54 +02:00
- **`CanBlockIsFalse`** - only run the test when the [[CanBlock]] property of the [Agent Record ](https://tc39.github.io/ecma262/#sec-agents ) executing the test file is `false`
- **`CanBlockIsTrue`** - only run the test when the [[CanBlock]] property of the [Agent Record ](https://tc39.github.io/ecma262/#sec-agents ) executing the test file is `true`
2014-08-22 17:47:18 +02:00
2015-05-25 23:51:55 +02:00
#### features
**features**: [list]
2014-07-21 17:39:10 +02:00
2017-10-17 22:38:39 +02:00
Some tests require the use of language features that are not directly described by the test file's location in the directory structure. These features should be specified with this tag. See the `features.txt` file for a complete list of available values. This tag is required for new tests written for new features, but contributions will not be "blocked" if the tag is missing from frontmatter. The committing maintainer is required to ensure that the tag is present and contains the correct feature names; this can be done in an follow up commit.
2014-07-30 00:27:20 +02:00
2017-10-06 22:09:39 +02:00
#### es5id
**es5id**: [es5-test-id]
**Deprecated.**
This tag identifies the section number from the portion of the ECMAScript 5.1 or ECMAScript 3 standard that is tested by this test. It was automatically generated for tests that were originally written for the ES5 (or earlier) version of the test suite and are now part of the ES6 version. You can use the es5id to discover the relevant portion of the ECMAScript standard by looking up the section number in [previous publications of the specification ](https://www.ecma-international.org/publications/standards/Ecma-262-arch.htm ). Unfortunately, there is no way to identify which version of ECMAScript (specifically, 3 or 5.1) without looking up the section number and deciding whether it covers the observable in the test.
Make Atomics tests more resilient against intermittent failures (#1617)
Also fixes multiple issues and bugs in various Atomics tests.
CONTRIBUTING.md and INTERPRETING.md:
- Add missing description for `CanBlockIsFalse` flag and also introduce `CanBlockIsTrue` flag which is needed for some tests (see below).
harness/atomicsHelper.js:
- Updated `$262.agent.waitUntil` to clarify it can also be called with BigInt64Array objects.
- Added `$262.agent.timeouts` constants to unify the various, inconsistently used timeouts in the Atomics tests. Each timeout constant has a brief description and a usage example to clarify when it should or should not be used. The default values were tested in SpiderMonkey (locally and CI) under various system load levels and should hopefully also be valid for other engines.
- Added `$262.agent.tryYield` to yield control from the main thread. The default implementation simply calls `$262.agent.sleep`, but test262 hosts could theoretically provide a different implementation.
- Added `$262.agent.trySleep` to replace direct calls to `$262.agent.sleep` from the main thread. Motivation for this function: Some test262 hosts (like browsers) may be able to pause and sleep on the main thread, so they could provide their implementation for `trySleep` which performs a busy-wait or something do nothing.
harness/testAtomics.js:
- Replace `let`, `arrow-functions`, and `for-of` loops to avoid creating extremely long `features` lists in tests using this helper. Removed `Map`, `WeakMap`, `Set`, and `WeakSet` (these weren't even listed in features.yaml) for the same reason and added the missing `Symbol` entry for this file to features.yaml.
- Updated all files including "harness/testAtomics.js" to match the new feature requirements.
test/built-ins/Atomics/store/good-views.js and test/built-ins/Atomics/store/bigint/good-views.js:
- Replace `arrow-functions` and `for-of` loops with ES5 alternatives.
test/built-ins/Atomics/wait/bigint/*.js and test/built-ins/Atomics/wake/bigint/*.js
- Some tests were using `BigInt64Array.BYTES_PER_ELEMENT * 8` for the SharedArrayBuffer length, but their non-BigInt counterparts are using `Int32Array.BYTES_PER_ELEMENT * 4`. For consistency and to make it easier to compare the BigInt against the non-BigInt versions, I've changed it to `BigInt64Array.BYTES_PER_ELEMENT * 4`.
- Also aligned formatting and statement placement when they differed between the non-BigInt and the BigInt version of a test file. (I've diffed some of the non-BigInt and BigInt files against each other and different formatting was a nuisance.)
Test files using `$262.agent.monotonicNow()`:
- Moved `$262.agent.report()` calls outside of the block of code measured `$262.agent.monotonicNow()` to avoid measuring how long it takes to execute `$262.agent.report()`.
- Without this change some tests failed intermittently in certain test configurations in SpiderMonkey. For example with the flags `--ion-eager -- ion-offthread-compile=off` which forces early Ion compilation on the main thread. The `$262.agent.report()` implementation in the SpiderMonkey test262 host embedding uses a for-loop which was forcefully Ion compiled under these settings. And because Ion compilation can take some time, the test case ran longer than `$262.agent.MAX_TIME_EPSILON` which lead to intermittent failures.
Test files using `CanBlockIsFalse` / `CanBlockIsTrue`:
- Some of these tests actually expected that the main thread can wait and [[CanBlock]] is `true` for the agent record executing the test. Therefore I've added a new `CanBlockIsTrue` flag and replaced the flags where needed.
test/built-ins/Atomics/wait/**/*.js and test/built-ins/Atomics/wake/**/*.js:
- Use an atomic counter `RUNNING` in more tests to have better control when a worker agent was actually started.
- Replace the various `$262.agent.sleep(/* Sleep X ms to ensure worker actually sleeps */)` calls with the new `$262.agent.tryYield()` function. This `X` was sometimes as low as 10 milliseconds, which is definitely too short for CI systems under heavy load (observed by intermittent CI failures for SpiderMonkey) and sometimes as high as 500 milliseconds, which is probably much longer than needed even when the system is under heavy load.
- Removed duplicate strings in assertion messages, presumably from copy-pasting the messages between different files.
- Removed extra empty lines at the end of multiple files.
test/built-ins/Atomics/wake/bad-range.js, test/built-ins/Atomics/wake/bigint/bad-range.js, and test/built-ins/Atomics/wait/bigint/non-bigint64-typedarray-throws.js:
- Removed unnecessary `features` and `includes` from this file.
test/built-ins/Atomics/wait/waiterlist-order-of-operations-is-fifo.js and test/built-ins/Atomics/wait/bigint/waiterlist-order-of-operations-is-fifo.js:
- The test was actually broken and didn't test what it said it does. This probably explains #1530.
- The test wants to ensure the waiterlist is implemented as a FIFO structure. This requires that the waiting agents all wait on the same index position, because the waiterlists are defined by each index. But if the agents wait on different indices, each agent is inserted into a different FIFO structure and therefore we can't observe any FIFO ordering between the agents when they're woken up.
- All this requires a bit of synchronization between the main agent and the waiting agent, I hope the added comments help to review these changes.
test/built-ins/Atomics/wait/good-views.js:
- The agent sends multiple reports to the main agent, but only the first one was read.
- This error was introduced during previous refactorings. I've changed it to back to use a while-loop as used in the first version of this file.
test/built-ins/Atomics/wait/no-spurious-wakeup-no-operation.js:
- Only the BigInt version of this test was present, copied it so we also get code coverage for the non-BigInt case.
test/built-ins/Atomics/wait/waiterlist-block-indexedposition-wake.js:
- Added extra while loops to avoid intermittent failures when the agent worker haven't started to wait.
- This should help to avoid some of the intermittent failures we saw for SpiderMonkey.
test/built-ins/Atomics/wake/wake-all-on-loc.js and test/built-ins/Atomics/wake/bigint/wake-all-on-loc.js:
- This test was also no longer after previous refactoring sessions.
- The "B" agent only waited for 10 milliseconds, which made it likely that it already timed out before the main agent was able to call `Atomics.wake`, which in turn rendered the test useless, because the test case wants to ensure that `Atomics.wake` cannot wake "B". But if "B" was already timed out, it can trivially not be woken by `Atomics.wake`.
- Added some safety measure to catch the case when "B" timed out before `Atomics.wake` was called and made it a test error if that happens.
test/built-ins/Atomics/wake/count-defaults-to-infinity-missing.js, test/built-ins/Atomics/wake/count-defaults-to-infinity-undefined.js, and test/built-ins/Atomics/wake/undefined-index-defaults-to-zero.js:
- Changed the `$262.agent.start()` calls to use a for-loop to avoid code duplication.
- (Forgot to undo the code formatting around `assert.sameValue`, still need to change it back.)
- Also more while-loops around `Atomics.wake`.
- These changes should fix #1529 and #1566.
test/built-ins/Atomics/wake/wake-all.js:
- Removed "B" worker agent.
- Without this removal the test case would be exactly equal to test/built-ins/Atomics/wake/wake-all-on-loc.js.
test/built-ins/Atomics/wake/wake-in-order-one-time.js:
- Add for-loops to avoid code duplication and make the test more readable.
- Make the `Atomics.wake` assertion messages unique by adding the current loop counter.
- Add `$262.agent.tryYield()` to give the worker agents enough time to actually start waiting.
test/built-ins/Atomics/wake/wake-in-order.js:
- Removed the outer loop `attempt < 10` because it uses `$262.agent` in a way currently not required to work. And which actually also doesn't work in SpiderMonkey's implementation of `$262.agent`.
- According to INTERPRETING.md `$262.agent.broadcast()` broadcasts its message to all agents ever started by `$262.agent.start()` and then blocks until all agents received the broadcast. It is not required that the agents started by the first `$262.agent.broadcast()` call will all be disabled/destroyed/whatever when the second `$262.agent.broadcast()` call occurs, which then means the second `$262.agent.broadcast()` call still tries to reach the agents started in the first loop iteration, but these may no longer accept broadcasts and therefore won't acknowledge they've received the broadcast. Which then means the second `$262.agent.broadcast()` call waits forever.
2018-07-03 18:34:54 +02:00
Read the [Test262 Technical Rationale Report ](https://github.com/tc39/test262/wiki/Test262-Technical-Rationale-Report,-October-2017#specification-reference-ids ) for reasoning behind deprecation.
2017-10-06 22:09:39 +02:00
#### es6id
**es6id**: [es6-test-id]
**Deprecated.**
2017-12-28 23:03:56 +01:00
This tag identifies the section number from the portion of the ES6 standard that is tested by this test _at the time the test was written_ . The es6ids might not correspond to the correction section numbers in the ES6 (or later) specification because routine edits to the specification will change section numbering. For this reason, only the esid is required for new tests.
2017-10-06 22:09:39 +02:00
Make Atomics tests more resilient against intermittent failures (#1617)
Also fixes multiple issues and bugs in various Atomics tests.
CONTRIBUTING.md and INTERPRETING.md:
- Add missing description for `CanBlockIsFalse` flag and also introduce `CanBlockIsTrue` flag which is needed for some tests (see below).
harness/atomicsHelper.js:
- Updated `$262.agent.waitUntil` to clarify it can also be called with BigInt64Array objects.
- Added `$262.agent.timeouts` constants to unify the various, inconsistently used timeouts in the Atomics tests. Each timeout constant has a brief description and a usage example to clarify when it should or should not be used. The default values were tested in SpiderMonkey (locally and CI) under various system load levels and should hopefully also be valid for other engines.
- Added `$262.agent.tryYield` to yield control from the main thread. The default implementation simply calls `$262.agent.sleep`, but test262 hosts could theoretically provide a different implementation.
- Added `$262.agent.trySleep` to replace direct calls to `$262.agent.sleep` from the main thread. Motivation for this function: Some test262 hosts (like browsers) may be able to pause and sleep on the main thread, so they could provide their implementation for `trySleep` which performs a busy-wait or something do nothing.
harness/testAtomics.js:
- Replace `let`, `arrow-functions`, and `for-of` loops to avoid creating extremely long `features` lists in tests using this helper. Removed `Map`, `WeakMap`, `Set`, and `WeakSet` (these weren't even listed in features.yaml) for the same reason and added the missing `Symbol` entry for this file to features.yaml.
- Updated all files including "harness/testAtomics.js" to match the new feature requirements.
test/built-ins/Atomics/store/good-views.js and test/built-ins/Atomics/store/bigint/good-views.js:
- Replace `arrow-functions` and `for-of` loops with ES5 alternatives.
test/built-ins/Atomics/wait/bigint/*.js and test/built-ins/Atomics/wake/bigint/*.js
- Some tests were using `BigInt64Array.BYTES_PER_ELEMENT * 8` for the SharedArrayBuffer length, but their non-BigInt counterparts are using `Int32Array.BYTES_PER_ELEMENT * 4`. For consistency and to make it easier to compare the BigInt against the non-BigInt versions, I've changed it to `BigInt64Array.BYTES_PER_ELEMENT * 4`.
- Also aligned formatting and statement placement when they differed between the non-BigInt and the BigInt version of a test file. (I've diffed some of the non-BigInt and BigInt files against each other and different formatting was a nuisance.)
Test files using `$262.agent.monotonicNow()`:
- Moved `$262.agent.report()` calls outside of the block of code measured `$262.agent.monotonicNow()` to avoid measuring how long it takes to execute `$262.agent.report()`.
- Without this change some tests failed intermittently in certain test configurations in SpiderMonkey. For example with the flags `--ion-eager -- ion-offthread-compile=off` which forces early Ion compilation on the main thread. The `$262.agent.report()` implementation in the SpiderMonkey test262 host embedding uses a for-loop which was forcefully Ion compiled under these settings. And because Ion compilation can take some time, the test case ran longer than `$262.agent.MAX_TIME_EPSILON` which lead to intermittent failures.
Test files using `CanBlockIsFalse` / `CanBlockIsTrue`:
- Some of these tests actually expected that the main thread can wait and [[CanBlock]] is `true` for the agent record executing the test. Therefore I've added a new `CanBlockIsTrue` flag and replaced the flags where needed.
test/built-ins/Atomics/wait/**/*.js and test/built-ins/Atomics/wake/**/*.js:
- Use an atomic counter `RUNNING` in more tests to have better control when a worker agent was actually started.
- Replace the various `$262.agent.sleep(/* Sleep X ms to ensure worker actually sleeps */)` calls with the new `$262.agent.tryYield()` function. This `X` was sometimes as low as 10 milliseconds, which is definitely too short for CI systems under heavy load (observed by intermittent CI failures for SpiderMonkey) and sometimes as high as 500 milliseconds, which is probably much longer than needed even when the system is under heavy load.
- Removed duplicate strings in assertion messages, presumably from copy-pasting the messages between different files.
- Removed extra empty lines at the end of multiple files.
test/built-ins/Atomics/wake/bad-range.js, test/built-ins/Atomics/wake/bigint/bad-range.js, and test/built-ins/Atomics/wait/bigint/non-bigint64-typedarray-throws.js:
- Removed unnecessary `features` and `includes` from this file.
test/built-ins/Atomics/wait/waiterlist-order-of-operations-is-fifo.js and test/built-ins/Atomics/wait/bigint/waiterlist-order-of-operations-is-fifo.js:
- The test was actually broken and didn't test what it said it does. This probably explains #1530.
- The test wants to ensure the waiterlist is implemented as a FIFO structure. This requires that the waiting agents all wait on the same index position, because the waiterlists are defined by each index. But if the agents wait on different indices, each agent is inserted into a different FIFO structure and therefore we can't observe any FIFO ordering between the agents when they're woken up.
- All this requires a bit of synchronization between the main agent and the waiting agent, I hope the added comments help to review these changes.
test/built-ins/Atomics/wait/good-views.js:
- The agent sends multiple reports to the main agent, but only the first one was read.
- This error was introduced during previous refactorings. I've changed it to back to use a while-loop as used in the first version of this file.
test/built-ins/Atomics/wait/no-spurious-wakeup-no-operation.js:
- Only the BigInt version of this test was present, copied it so we also get code coverage for the non-BigInt case.
test/built-ins/Atomics/wait/waiterlist-block-indexedposition-wake.js:
- Added extra while loops to avoid intermittent failures when the agent worker haven't started to wait.
- This should help to avoid some of the intermittent failures we saw for SpiderMonkey.
test/built-ins/Atomics/wake/wake-all-on-loc.js and test/built-ins/Atomics/wake/bigint/wake-all-on-loc.js:
- This test was also no longer after previous refactoring sessions.
- The "B" agent only waited for 10 milliseconds, which made it likely that it already timed out before the main agent was able to call `Atomics.wake`, which in turn rendered the test useless, because the test case wants to ensure that `Atomics.wake` cannot wake "B". But if "B" was already timed out, it can trivially not be woken by `Atomics.wake`.
- Added some safety measure to catch the case when "B" timed out before `Atomics.wake` was called and made it a test error if that happens.
test/built-ins/Atomics/wake/count-defaults-to-infinity-missing.js, test/built-ins/Atomics/wake/count-defaults-to-infinity-undefined.js, and test/built-ins/Atomics/wake/undefined-index-defaults-to-zero.js:
- Changed the `$262.agent.start()` calls to use a for-loop to avoid code duplication.
- (Forgot to undo the code formatting around `assert.sameValue`, still need to change it back.)
- Also more while-loops around `Atomics.wake`.
- These changes should fix #1529 and #1566.
test/built-ins/Atomics/wake/wake-all.js:
- Removed "B" worker agent.
- Without this removal the test case would be exactly equal to test/built-ins/Atomics/wake/wake-all-on-loc.js.
test/built-ins/Atomics/wake/wake-in-order-one-time.js:
- Add for-loops to avoid code duplication and make the test more readable.
- Make the `Atomics.wake` assertion messages unique by adding the current loop counter.
- Add `$262.agent.tryYield()` to give the worker agents enough time to actually start waiting.
test/built-ins/Atomics/wake/wake-in-order.js:
- Removed the outer loop `attempt < 10` because it uses `$262.agent` in a way currently not required to work. And which actually also doesn't work in SpiderMonkey's implementation of `$262.agent`.
- According to INTERPRETING.md `$262.agent.broadcast()` broadcasts its message to all agents ever started by `$262.agent.start()` and then blocks until all agents received the broadcast. It is not required that the agents started by the first `$262.agent.broadcast()` call will all be disabled/destroyed/whatever when the second `$262.agent.broadcast()` call occurs, which then means the second `$262.agent.broadcast()` call still tries to reach the agents started in the first loop iteration, but these may no longer accept broadcasts and therefore won't acknowledge they've received the broadcast. Which then means the second `$262.agent.broadcast()` call waits forever.
2018-07-03 18:34:54 +02:00
Read the [Test262 Technical Rationale Report ](https://github.com/tc39/test262/wiki/Test262-Technical-Rationale-Report,-October-2017#specification-reference-ids ) for reasoning behind deprecation.
2017-10-06 22:09:39 +02:00
2014-08-22 17:47:18 +02:00
## Test Environment
2014-07-30 00:27:20 +02:00
Each test case is run in a fresh JavaScript environment; in a browser, this will be a new `IFRAME` ; for a console runner, this will be a new process. The test harness code is loaded before the test is run. The test harness defines the following helper functions:
Function | Purpose
---------|--------
Test262Error(message) | constructor for an error object that indicates a test failure
2015-05-25 23:51:55 +02:00
$DONE(arg) | see Writing Asynchronous Tests, below
assert(value, message) | throw a new Test262Error instance if the specified value is not strictly equal to the JavaScript `true` value; accepts an optional string message for use in creating the error
2017-04-14 18:19:39 +02:00
assert.sameValue(actual, expected, message) | throw a new Test262Error instance if the first two arguments are not [the same value ](https://tc39.github.io/ecma262/#sec-samevalue ); accepts an optional string message for use in creating the error
assert.notSameValue(actual, unexpected, message) | throw a new Test262Error instance if the first two arguments are [the same value ](https://tc39.github.io/ecma262/#sec-samevalue ); accepts an optional string message for use in creating the error
2017-05-01 18:29:52 +02:00
assert.throws(expectedErrorConstructor, fn, message) | throw a new Test262Error instance if the provided function does not throw an error, or if the constructor of the value thrown does not match the provided constructor
2017-05-01 19:22:49 +02:00
assert.throws.early(expectedErrorConstructor, code) | throw a new Test262Error instance if the provided code does not throw an early error, or if the constructor of the value thrown does not match the provided constructor. This assertion catches only errors that will be parsed through `Function(code)` .
2018-05-10 00:42:56 +02:00
$ERROR(message) | construct a Test262Error object and throw it < br > **DEPRECATED** -- Do not use in new tests. Use `assert` , `assert.*` , or `throw new Test262Error` instead.
2015-05-25 23:51:55 +02:00
2014-07-30 00:27:20 +02:00
```
/// error class
function Test262Error(message) {
//[omitted body]
}
```
2014-08-22 17:47:18 +02:00
## Handling Errors and Negative Test Cases
2014-07-30 00:27:20 +02:00
2017-04-29 21:33:06 +02:00
Expectations for **parsing errors** should be declared using [the `negative` frontmatter flag ](#negative ). They must also include the exact `throw` statement specified in this example (in order to guarantee that implementations do not execute the code):
2014-07-30 00:27:20 +02:00
2014-07-04 02:13:42 +02:00
```javascript
2014-07-30 00:27:20 +02:00
/*---
2016-03-13 18:26:10 +01:00
negative:
2017-12-03 06:06:42 +01:00
phase: parse
2016-03-13 18:26:10 +01:00
type: SyntaxError
2014-07-30 00:27:20 +02:00
---*/
2014-07-04 02:13:42 +02:00
2017-04-29 21:33:06 +02:00
throw "Test262: This statement should not be evaluated.";
2015-05-25 23:51:55 +02:00
var var = var;
2014-07-04 02:13:42 +02:00
```
2015-05-25 23:51:55 +02:00
Expectations for **runtime errors** should be defined using the `assert.throws` method and the appropriate JavaScript Error constructor function:
2014-08-22 17:47:18 +02:00
2014-07-04 02:13:42 +02:00
```javascript
2016-03-22 14:51:05 +01:00
assert.throws(TypeError, function() {
null(); // expect this statement to throw a TypeError
2015-05-25 23:51:55 +02:00
});
2014-07-04 02:13:42 +02:00
```
2014-07-21 17:39:10 +02:00
2014-08-22 17:47:18 +02:00
## Writing Asynchronous Tests
2014-07-21 17:39:10 +02:00
2016-02-12 18:59:51 +01:00
An asynchronous test is any test that include the `async` frontmatter flag. When executing such tests, the runner expects that the global `$DONE()` function will be called to signal test completion.
2014-07-21 17:39:10 +02:00
* If the argument to `$DONE` is omitted, is `undefined` , or is any other falsy value, the test is considered to have passed.
* If the argument to `$DONE` is a truthy value, the test is considered to have failed and the argument is displayed as the failure reason.
A common idiom when writing asynchronous tests is the following:
```js
var p = new Promise(function () { /* some test code */ });
p.then(function checkAssertions(arg) {
2017-04-14 18:19:39 +02:00
if (!expected_condition) {
2017-10-06 22:09:39 +02:00
throw new Test262Error("failure message");
2017-04-14 18:19:39 +02:00
}
2014-07-21 17:39:10 +02:00
}).then($DONE, $DONE);
```
2014-07-30 00:27:20 +02:00
Function `checkAssertions` implicitly returns `undefined` if the expected condition is observed. The return value of function `checkAssertions` is then used to asynchronously invoke the first function of the final `then` call, resulting in a call to `$DONE(undefined)` , which signals a passing test.
2014-07-21 17:39:10 +02:00
2017-10-06 22:09:39 +02:00
If the expected condition is not observed, function `checkAssertions` throws a `Test262Error` . This is caught by the Promise and then used to asynchronously invoke the second function in the call -- which is also `$DONE` -- resulting in a call to `$DONE(error_object)` , which signals a failing test.
2014-07-21 17:39:10 +02:00
2014-08-22 17:47:18 +02:00
### Checking Exception Type and Message in Asynchronous Tests
2014-07-21 17:39:10 +02:00
This idiom can be extended to check for specific exception types or messages:
```js
p.then(function () {
2017-04-14 18:19:39 +02:00
// some code that is expected to throw a TypeError
2014-07-21 17:39:10 +02:00
2017-04-14 18:19:39 +02:00
return "Expected exception to be thrown";
2014-07-21 17:39:10 +02:00
}).then($DONE, function (e) {
2017-04-14 18:19:39 +02:00
if (!(e instanceof TypeError)) {
2017-10-06 22:09:39 +02:00
throw new Test262Error("Expected TypeError but got " + e);
2017-04-14 18:19:39 +02:00
}
2014-07-21 17:39:10 +02:00
2017-04-14 18:19:39 +02:00
if (!/expected message/.test(e.message)) {
2017-10-06 22:09:39 +02:00
throw new Test262Error("Expected message to contain 'expected message' but found " + e.message);
2017-04-14 18:19:39 +02:00
}
2014-07-21 17:39:10 +02:00
}).then($DONE, $DONE);
```
2015-05-25 23:51:55 +02:00
As above, exceptions that are thrown from a `then` clause are passed to a later `$DONE` function and reported asynchronously.
2016-03-16 18:58:29 +01:00
2017-05-01 18:04:05 +02:00
## Linting
2017-12-03 05:09:24 +01:00
Some of the expectations documented here are enforced via a "linting" script. This script is used to validate patches automatically at submission time, but it may also be invoked locally. To do so, first install the required Python packages via the following command:
pip install --requirement tools/lint/requirements.txt
Then invoke the following command:
2017-05-01 18:04:05 +02:00
python tools/lint/lint.py --whitelist lint.whitelist [paths to tests]
...where `[paths to tests]` is a list of one or more paths to test files or directories containing test files.
In some cases, it may be necessary for a test to intentionally violate the rules enforced by the linting tool. Such violations can be allowed by including the path of the test(s) in the `lint.whitelist` file. Each path must appear on a dedicated line in that file, and a space-separated list of rules to ignore must follow each path. Lines beginning with the pound sign (`#`) will be ignored. For example:
# This file documents authorship information and is not itself a test
test/built-ins/Simd/AUTHORS FRONTMATTER LICENSE
2016-03-16 18:58:29 +01:00
## Procedurally-generated tests
Some language features are expressed through a number of distinct syntactic forms. Test262 maintains these tests as a set of "test cases" and "test templates" in order to ensure equivalent coverage across all forms. The sub-directories within the `src/` directory describe the various language features that benefit from this approach.
Test cases and test templates specify meta-data using the same YAML frontmatter pattern as so-called "static" (i.e. non-generated) tests. The expected attributes differ between test cases and test templates:
2017-10-17 18:57:44 +02:00
### test cases (`*.case`)
Field | Description
------|-------------
`template` | name of the sub-directory to locate templates for this test
`desc` | see the frontmatter definition of the "desc" field. The generated test will have a have final "desc" value which is this text appended with the test template's "name" field in parentheses.
2017-10-21 02:46:04 +02:00
`info` | see the frontmatter definition of the "info" field. The generated test will have a have final "info" value which is this text concatenated at the end of the test templates's "info" text.
2017-10-17 18:57:44 +02:00
`features` | see the frontmatter definition of the "features" field. The generated test will have a final feature list in combination with the template's feature field.
### test templates (`*.template`)
Field | Description
------|-------------
`path` | location within the published test hierarchy to output files created from this template. This path will be ended with the name of the test case file. If path is "/test/language/template1-" and the test case is "cast1.js", the final location of the file will be "/test/language/template1-case1.js"
`name` | human-readable name of the syntactic form described by this template. This text will be appended, in parentheses, to the end of the test cases `desc` field.
`esid` | see the frontmatter definition of the "info" tag.
2017-10-21 02:46:04 +02:00
`info` | see the frontmatter definition of the "info" tag. The generated test will have a have final "info" value which is this text appended with the test cases's "info" text.
2017-10-17 18:57:44 +02:00
`features` | see the frontmatter definition of the "features" field. The generated test will have a final feature list in combination with the test case's feature field.
any other valid frontmatter field | see the frontmatter definitions.
2016-03-16 18:58:29 +01:00
2017-12-03 05:09:24 +01:00
Generated files are managed using the `make.py` Python script located in the root of this repository. To use it, first install the required Python packages via the following command:
pip install --requirement tools/generation/requirements.txt
2016-03-16 18:58:29 +01:00
2017-12-03 05:09:24 +01:00
And then issue the following command to create files:
2016-03-16 18:58:29 +01:00
make.py
To remove all generated files:
make.py clean
The executable located at `tools/generation/generator.py` offers additional control over the generation procedure.
./tools/generation/generator.py --help
2017-04-10 21:31:19 +02:00
Tests expressed with this convention are built automatically following the source files' acceptance into the project. Patches should **not** include assets built from these sources.