Cleanup for a static constructor - node.js

So I'm trying to write some unit tests for my library. But when running jest I get the dreaded:
Jest did not exit one second after the test run has completed.
This usually means that there are asynchronous operations that weren't
stopped in your tests. Consider running Jest with `--detectOpenHandles`
to troubleshoot this issue.
I've tracked this down to a specific class I've written that has a static constructor block. In that block I start a setInterval, or a setTimeout function. Either way I try it jest acts the same way.
So I know Javascript doesn't have destructors for some unknown reason (GC isn't a valid reason. Lots of languages with GCs have destructors). How to I clean up and stop the setInterval when the code is ready to shutdown?
Option 1:
class A {
static #blink = false;
static #blinker;
static {
A.#blinker = setInterval(() => {
A.#blink = !A.#blink;
}, 500);
}
}
Option 2:
class B {
static #blink = false;
static #blinker = null;
static {
B.#doBlink();
}
static #doBlink() {
B.#blink = !B.#blink;
#blinker = setTimeout(() => B.#doBlink(), 500);
}
}

There are numerous choices:
Add a .shutdown() or .close() method to your object and call that method when you're done with the object and want the timers to stop.
call .unref() on the timers and then they won't keep nodejs from shutting down naturally - they won't count as an unfinished asynchronous operation when the nodejs event loop is detecting whether anything is still running or not.
Using the logic of your code (which you don't show), figure out when the timers are no longer needed and shut down the setTimeout() or setInterval() naturally based on those conditions.

Related

Kotlin: Why isn't job.invokeOnCompletion() block running on main thread?

In my Android application I have code that should run periodically in its own coroutine and should be cancelable.
for this I have the following functions:
startJob(): Initializes the job, sets up invokeOnCompletion() and starts the work loop in the respective scope
private fun startJob() {
if (::myJob.isInitialized && myJob.isActive) {
return
}
myJob= Job()
myJob.invokeOnCompletion {
it?.message.let {
var msg = it
if (msg.isNullOrBlank()) {
msg = "Job stopped. Reason unknown"
}
myJobCompleted(msg)
}
}
CoroutineScope(Dispatchers.IO + myJob).launch {
workloop()
}
}
workloop(): The main work loop. Do some work in a loop with a set delay in each iteration:
private suspend fun workloop() {
while (true) {
// doing some stuff here
delay(setDelayInMilliseconds)
}
}
myJobCompleted: do some finalizing. For now simply log a message for testing.
private fun myJobCompleted(msg: String) {
try {
mainActivityReference.logToGUI(msg)
}
catch (e:Exception){
println("debug: " + e.message)
}
}
Running this and calling myJob.Cancel() will throw the following exception in myJobCompleted():
debug: Only the original thread that created a view hierarchy can touch its views.
I'm curious as to why this code isn't running on the main thread, since startJob() IS called from the main thread?
Furthermore: is there a option similar to using a CancellationTokenSource in c#, where the job is not immediately cancelled, but a cancellation request can be checked each iteration of the while loop?
Immediately breaking off the job, regardless of what it is doing (although it will pretty much always be waiting for the delay on cancellation) doesn't seem like a good idea to me.
It is not the contract of Job.invokeOnCompletion to run on the same thread where Job is created. Moreover, such a contract would be impossible to implement.
You can't expect an arbitrary piece of code to run on an arbitrary thread, just because there was some earlier method invocation on that thread. The ability of the Android main GUI thread to execute code submitted from the outside is special, and involves the existence a top-level event loop.
In the world of coroutines, what controls thread assignment is the coroutine context, while clearly you are outside of any context when creating the job. So the way to fix it is to explicitly launch(Dispatchers.Main) a coroutine from within invokeOnCompletion.
About you question on cancellation, you can use withContext(NonCancellable) to surround the part of code you want to protect from cancellation.

Node.js Spawning multiple threads within a class method

How can I run a single method multiple times multi-threaded when called as a method of a class?
At first I tried to use the cluster module, but I realize it just re-runs the whole process from the start, rightfully so.
How can I achieve something like what's outlined below?
I want a class's method to spawn n processes, and when the parallel tasks are completed, I can resolve a promise which the method returns.
The problem with the code below is that calling cluster.fork() will fork index.js process.
index.js
const Person = require('./Person.js');
var Mary = new Person('Mary');
Mary.run(5).then(() => {...});
console.log('I should only run once, but I am called 5 times too many');
Person.js
const cluster = require('cluster');
class Person{
run(distance){
var completed = 0;
return new Promise((resolve, reject) => {
for(var i = 0; i < distance; i++) {
// run a separate process for each
cluster.fork().send(i).on('message', message => {
if (message === 'completed') { ++completed; }
if (completed === distance) { resolve(); }
});
}
});
}
}
I think the short answer is impossible. It's even worse - this has nothing to do with js. To multi (process or thread) in your particular problem you will essentially need a copy of the object in every thread, since it needs (maybe) access to fields - in this case you would need to either initialize it in every thread or share memory. That last one I don't think is provided in cluster, and not trivial in other languages in every use case.
If the calculation is independent of the Person I suggest you extract it, and use the usual (in index.js):
if(cluster.isWorker) {
//Use the i for calculation
} else {
//Create Person, then fork children in for loop
}
You then collect the results and change the Person as needed. You will be copying index.js, but this is standard and you only run what you need.
The problem is if results are dependent on Person. If these are constant for all i you can still send them to your forks independently. Otherwise what you have is the only way to fork. In general forking in cluster is not meant for methods, but for the app itself, which is the standard forking behavior.
Another solution
Following your comment, I suggest you checkout child_process.execFile or child_process.exec on same file.
This way you can spawn a totally independent process on the fly. Now instead of calling cluster.fork you call execFile. You can use either the exit code or stdout as return values (stderr etc.). Promise is now replaced with:
var results = []
for(var i = 0; i < distance; i++) {
// run a separate process for each
results.push(child_process.execFile().child.execFile('node', 'mymethod.js`,i]));
}
//... catch the exit event from all results or return a callback using results.
Inside mymethod.js Have your code that takes i and returns what you want either in the exit code or through stdout, both properties of the returned child_process. This is a bit un-node.js-y since you're waiting on asynchronous calls, but you're requirements are non standard. Since I'm not sure how you use this perhaps returning a callback with the array is a better idea.

How to forcibly keep a Node.js process from terminating?

TL;DR
What is the best way to forcibly keep a Node.js process running, i.e., keep its event loop from running empty and hence keeping the process from terminating? The best solution I could come up with was this:
const SOME_HUGE_INTERVAL = 1 << 30;
setInterval(() => {}, SOME_HUGE_INTERVAL);
Which will keep an interval running without causing too much disturbance if you keep the interval period long enough.
Is there a better way to do it?
Long version of the question
I have a Node.js script using Edge.js to register a callback function so that it can be called from inside a DLL in .NET. This function will be called 1 time per second, sending a simple sequence number that should be printed to the console.
The Edge.js part is fine, everything is working. My only problem is that my Node.js process executes its script and after that it runs out of events to process. With its event loop empty, it just terminates, ignoring the fact that it should've kept running to be able to receive callbacks from the DLL.
My Node.js script:
var
edge = require('edge');
var foo = edge.func({
assemblyFile: 'cs.dll',
typeName: 'cs.MyClass',
methodName: 'Foo'
});
// The callback function that will be called from C# code:
function callback(sequence) {
console.info('Sequence:', sequence);
}
// Register for a callback:
foo({ callback: callback }, true);
// My hack to keep the process alive:
setInterval(function() {}, 60000);
My C# code (the DLL):
public class MyClass
{
Func<object, Task<object>> Callback;
void Bar()
{
int sequence = 1;
while (true)
{
Callback(sequence++);
Thread.Sleep(1000);
}
}
public async Task<object> Foo(dynamic input)
{
// Receives the callback function that will be used:
Callback = (Func<object, Task<object>>)input.callback;
// Starts a new thread that will call back periodically:
(new Thread(Bar)).Start();
return new object { };
}
}
The only solution I could come up with was to register a timer with a long interval to call an empty function just to keep the scheduler busy and avoid getting the event loop empty so that the process keeps running forever.
Is there any way to do this better than I did? I.e., keep the process running without having to use this kind of "hack"?
The simplest, least intrusive solution
I honestly think my approach is the least intrusive one:
setInterval(() => {}, 1 << 30);
This will set a harmless interval that will fire approximately once every 12 days, effectively doing nothing, but keeping the process running.
Originally, my solution used Number.POSITIVE_INFINITY as the period, so the timer would actually never fire, but this behavior was recently changed by the API and now it doesn't accept anything greater than 2147483647 (i.e., 2 ** 31 - 1). See docs here and here.
Comments on other solutions
For reference, here are the other two answers given so far:
Joe's (deleted since then, but perfectly valid):
require('net').createServer().listen();
Will create a "bogus listener", as he called it. A minor downside is that we'd allocate a port just for that.
Jacob's:
process.stdin.resume();
Or the equivalent:
process.stdin.on("data", () => {});
Puts stdin into "old" mode, a deprecated feature that is still present in Node.js for compatibility with scripts written prior to Node.js v0.10 (reference).
I'd advise against it. Not only it's deprecated, it also unnecessarily messes with stdin.
Use "old" Streams mode to listen for a standard input that will never come:
// Start reading from stdin so we don't exit.
process.stdin.resume();
Here is IFFE based on the accepted answer:
(function keepProcessRunning() {
setTimeout(keepProcessRunning, 1 << 30);
})();
and here is conditional exit:
let flag = true;
(function keepProcessRunning() {
setTimeout(() => flag && keepProcessRunning(), 1000);
})();
You could use a setTimeout(function() {""},1000000000000000000); command to keep your script alive without overload.
spin up a nice repl, node would do the same if it didn't receive an exit code anyway:
import("repl").then(repl=>
repl.start({prompt:"\x1b[31m"+process.versions.node+": \x1b[0m"}));
I'll throw another hack into the mix. Here's how to do it with Promise:
new Promise(_ => null);
Throw that at the bottom of your .js file and it should run forever.

Code coverage for async methods

When I analyse code coverage in Visual Studio 2012, any of the await lines in async methods are showing as not covered even though they are obviously executing since my tests are passing. The code coverage report says that the uncovered method is MoveNext, which is not present in my code (perhaps it's compiler-generated).
Is there a way to fix code coverage reporting for async methods?
Note:
I just ran coverage using NCover, and the coverage numbers make a lot more sense using that tool. As a workaround for now, I'll be switching to that.
This can happen most commonly if the operation you're awaiting is completed before it's awaited.
I recommend you test at least synchronous and asynchronous success situations, but it's also a good idea to test synchronous and asynchronous errors and cancellations.
The reason the code is not shown as being covered has to do with how async methods are implemented. The C# compiler actually translates the code in async methods into a class that implements a state machine, and transforms the original method into a stub that initialized and invokes that state machine. Since this code is generated in your assembly, it is included in the code coverage analysis.
If you use a task that is not complete at the time the code being covered is executing, the compiler-generated state machine hooks up a completion callback to resume when the task completes. This more completely exercises the state machine code, and results in complete code coverage (at least for statement-level code coverage tools).
A common way to get a task that is not complete at the moment, but will complete at some point is to use Task.Delay in your unit test. However, that is generally a poor option because the time delay is either too small (and results in unpredictable code coverage because sometimes the task is complete before the code being tests runs) or too large (unnecessarily slowing the tests down).
A better option is to use "await Task.Yield()". This will return immediately but invoke the continuation as soon as it is set.
Another option - though somewhat absurd - is to implement your own awaitable pattern that has the semantics of reporting incomplete until a continuation callback is hooked up, and then to immediately complete. This basically forces the state machine into the async path, providing the complete coverage.
To be sure, this is not a perfect solution. The most unfortunate aspect is that it requires modification to production code to address a limitation of a tool. I would much prefer that the code coverage tool ignore the portions of the async state machine that are generated by the compiler. But until that happens, there aren’t many options if you really want to try to get complete code coverage.
A more complete explanation of this hack can be found here: http://blogs.msdn.com/b/dwayneneed/archive/2014/11/17/code-coverage-with-async-await.aspx
There are situations where I don't care about testing the async nature of a method but just want to get rid of the partial code coverage. I use below extension method to avoid this and it works just fine for me.
Warning "Thread.Sleep" used here!
public static IReturnsResult<TClass> ReturnsAsyncDelayed<TClass, TResponse>(this ISetup<TClass, Task<TResponse>> setup, TResponse value) where TClass : class
{
var completionSource = new TaskCompletionSource<TResponse>();
Task.Run(() => { Thread.Sleep(200); completionSource.SetResult(value); });
return setup.Returns(completionSource.Task);
}
and the usage is similar to the Moq's ReturnsAsync Setup.
_sampleMock.Setup(s => s.SampleMethodAsync()).ReturnsAsyncDelayed(response);
I created a test runner that runs a block of code multiple times and varies the task that is delayed using a factory. This is great for testing the different paths through simple blocks of code. For more complex paths you may want to create a test per path.
[TestMethod]
public async Task ShouldTestAsync()
{
await AsyncTestRunner.RunTest(async taskFactory =>
{
this.apiRestClient.GetAsync<List<Item1>>(NullString).ReturnsForAnyArgs(taskFactory.Result(new List<Item1>()));
this.apiRestClient.GetAsync<List<Item2>>(NullString).ReturnsForAnyArgs(taskFactory.Result(new List<Item2>()));
var items = await this.apiController.GetAsync();
this.apiRestClient.Received().GetAsync<List<Item1>>(Url1).IgnoreAwait();
this.apiRestClient.Received().GetAsync<List<Item2>>(Url2).IgnoreAwait();
Assert.AreEqual(0, items.Count(), "Zero items should be returned.");
});
}
public static class AsyncTestRunner
{
public static async Task RunTest(Func<ITestTaskFactory, Task> test)
{
var testTaskFactory = new TestTaskFactory();
while (testTaskFactory.NextTestRun())
{
await test(testTaskFactory);
}
}
}
public class TestTaskFactory : ITestTaskFactory
{
public TestTaskFactory()
{
this.firstRun = true;
this.totalTasks = 0;
this.currentTestRun = -1; // Start at -1 so it will go to 0 for first run.
this.currentTaskNumber = 0;
}
public bool NextTestRun()
{
// Use final task number as total tasks.
this.totalTasks = this.currentTaskNumber;
// Always return has next as turn for for first run, and when we have not yet delayed all tasks.
// We need one more test run that tasks for if they all run sync.
var hasNext = this.firstRun || this.currentTestRun <= this.totalTasks;
// Go to next run so we know what task should be delayed,
// and then reset the current task number so we start over.
this.currentTestRun++;
this.currentTaskNumber = 0;
this.firstRun = false;
return hasNext;
}
public async Task<T> Result<T>(T value, int delayInMilliseconds = DefaultDelay)
{
if (this.TaskShouldBeDelayed())
{
await Task.Delay(delayInMilliseconds);
}
return value;
}
private bool TaskShouldBeDelayed()
{
var result = this.currentTaskNumber == this.currentTestRun - 1;
this.currentTaskNumber++;
return result;
}
public async Task VoidResult(int delayInMilliseconds = DefaultDelay)
{
// If the task number we are on matches the test run,
// make it delayed so we can cycle through them.
// Otherwise this task will be complete when it is reached.
if (this.TaskShouldBeDelayed())
{
await Task.Delay(delayInMilliseconds);
}
}
public async Task<T> FromResult<T>(T value, int delayInMilliseconds = DefaultDelay)
{
if (this.TaskShouldBeDelayed())
{
await Task.Delay(delayInMilliseconds);
}
return value;
}
}

How to optimize tests validating asynchronous code?

We are developing a WPF application using TDD. As we're already working on this solution for almost two years, we've written a huge bunch of tests (almost 2000 Unittests right now).
There are some classes, that need to implement functionality multithreaded and asynchronously. For example a communication-component that can both send and receive messages and parse them. The dependencies are always mocked using RhinoMocks.
Our Test-Methods targeting these classes look very similar, as following:
[TestMethod]
public void Method_Description_ExpectedResult(){
// Arrange
var myStub = MockRepository.GenerateStub<IMyStub>();
var target = new MyAsynchronousClass(myStub);
// Act
var target.Send("Foo");
Thread.Sleep(200);
//Assert
myStub.AssertWasCalled(x => x.Bar("Foo"));
}
As you can see, this test runs at least for 200 ms due to the Thread.Sleep(). We optimized the test replacing the AssertWasCalled with a active polling method, s.th. like this:
public static bool True(Func<bool> condition, int times, int waitTime)
{
for (var i = 0; i < times; i++)
{
if (condition())
return true;
Thread.Sleep(waitTime);
}
return condition();
}
We can now use this WaitFor.True(...) Method by changing the AssertWasCalled to:
var fooTriggered = false;
myStub.Stub(x => x.Bar("Foo")).Do((Action)(() => fooTriggered = true)));
WaitFor.True(() => fooTriggered, 20, 20);
Assert.IsTrue(fooTriggered);
This construct will terminate earlier if the condition matches, but anyway - this takes too long for us. Running all of our 2000 Tests need about 5 Minutes (building and running them).
Is there any smart trick how we could optimize code like this?
You can use a monitor. I'm making this up so please excuse me if it isn't quite compiling, but it'll look something like:
[TestMethod]
public void Method_Description_ExpectedResult(){
// Arrange
var waitingRoom = new object();
var myStub = MockRepository.GenerateStub<IMyStub>();
myStub.Setup(x => x.Bar("Foo")).Callback(x =>
{
Monitor.Enter(waitingRoom);
Monitor.Pulse(waitingRoom);
Monitor.Exit(waitingRoom);
}
var target = new MyAsynchronousClass(myStub);
// Act
Monitor.Enter(waitingRoom);
target.Send("Foo");
Monitor.Wait(waitingRoom);
Monitor.Exit(waitingRoom);
//Assert
myStub.AssertWasCalled(x => x.Bar("Foo"));
}
Code written within the Monitor can't run until it's free. The test will cause the acting thread to wait until Monitor.Wait has been called. Then the callback can enter and pulse the Monitor. The test then "wakes up", and once the callback has exited the monitor, it gets control back and exits too, allowing you to Assert.
The only thing I haven't covered is that if Bar("Foo") doesn't get called it will hang, so you might want to have a timer pulse the thread too.
You can create a class which does the complex monitoring bits for you if you use it a lot. This is one I wrote to deal with asynchronous checks in UI automation; adapting it for what you're doing might help you.

Resources