I'm trying to build an NES emulator using winit, which entails building a game loop which should run exactly 60 times per second.
At first, I used std::thread to create a separate thread where the game loop would run and wait 16 milliseconds before running again. This worked quite well, until I tried to compile the program again targeting WebAssembly. I then found out that both winit::window::Window and winit::event_loop::EventLoopProxy are not Send when targeting Wasm, and that std::thread::spawn panics in Wasm.
After some struggle, I decided to try to do the same thing using task::spawn_local from one of the main asynchronous runtimes. Ultimately, I went with async_std.
I'm not used to asynchronous programming, so I'm not even sure if what I'm trying to do could work.
My idea is to do something like this:
use winit::{window::WindowBuilder, event_loop::EventLoop};
use std::time::Duration;
fn main() {
let event_loop = EventLoop::new();
let _window = WindowBuilder::new()
.build(&event_loop);
async_std::task::spawn_local(async {
// game loop goes here
loop {
// [update game state]
// [update frame buffer]
// [send render event with EventLoopProxy]
async_std::task::sleep(Duration::from_millis(16)).await;
// ^ note: I'll be using a different sleep function with Wasm
}
});
event_loop.run(move |event, _, control_flow| {
control_flow.set_wait();
match event {
// ...
_ => ()
}
});
}
The problem with this approach is that the game loop will never run. If I'm not mistaken, some asynchronous code in the main thread would need to be blocked (by calling .await) for the runtime to poll other Futures, such as the one spawned by the spawn_local function. I can't do this easily, since event_loop.run is not asynchronous.
Having time to await other events shouldn't be a problem, since the control flow is set to wait.
Testing this on native code, nothing inside the game loop ever runs. Testing this on Wasm code (with wasm_timer::Delay as the sleep function), the game loop does run, but at a very low framerate and with long intervals of halting.
Having explained my situation, I would like to ask: is it possible to do what I'm trying to do, and if it is, how would I approach it? I will also accept answers telling me how I could try to do this differently, such as by using web workers.
Thanks in advance!
Related
using QtCreator to make a loftier interface to a sofware.
There is basically a set of buttons to tune and inputs, a start and stop job
buttons, my problem comes from an infinite loop that freezes the display so I came up with using fork() so that the loop have to compete with the main program instead of eating up the whole resources (no multithreading), but the program crashes spiting:
[xcb] Unknown sequence number while processing queue
[xcb] Most likely this is a multi-threaded client and XInitThreads has not
been called
[xcb] Aborting, sorry about that.
a.out: ../../src/xcb_io.c:274: poll_for_event: Assertion
`!xcb_xlib_threads_sequence_lost' failed.
the fonction calling the loop is called 'ON', 'OFF' is supposed to exit the forked process.
//in button 'ON' func
ps = fork();
if(getpid() == ps)
{
while(1)
{
strcpy(word, charset(minlength, maxlength, N));
ui->pass->setText(word);//operation on the display
....SNIP
}
}
//In button 'OFF' func
if(getpid() == ps)
exit(0);
I'm really asking what is the correct way of starting a while(1) and be able to break, exit, or return from it while not freezing the window using QT, thanks.
You crash probably here:
ui->pass->setText(word);//operation on the display
as in Qt, you can not change UI from non UI threads directly. Only from signals and slots mechanism.
The proper way to not freeze UI is obviously to compute lengthy operations in another thread.
You can achieve this in several ways:
One is by sub-classing QObject class to create 'worker object' which would perform all heavy operations. You create new QThread object that should live as long as you need your object. And use QObject::moveToThread method to move created object to new thread. To control your worker object you should send signals from object and call it's slots also via signal-slot mechanism. If you call them directly - they will be executed in caller thread (so do not perform stuff like worker->startHeavyJob(); in UI thread). Instead emit signal in UI (emit sigStartHeavyStuff();) and connect it to slot of your worker object (slotDoHeavyStuff();)
if you do not want to bother with this (if operation is pretty small)
- you can use QApplication::processEvents() to process events in UI event loop while going in your infinite while loop.
Another way is to use QtConcurrentRun framework to run function in separate thread which manages itself. Threads are taken from thread pool and are managed by Qt. This approach looks like something you want to do. Though you still will be able to access UI objects only through signals and slots.
I see one big issue in the presented code that is causing your freeze: You never let Qt process anything while you are in the loop. You need to allow Qt to run it's event loop. The easiest way is to use QApplication::processEvents() inside the loop.
I'm also not a fan of a while(1) loop for a couple of reasons. The first of which is that it can eat machine cycles waiting for things to happen. I doubt you really need to run the code as fast as possible, you could probably get away with some sleeping in your loop.
The other issue is that it is hard to break out. A cleaner approach would be something like this
void MyClass::on_pushButton_ON_clicked()
{
MyClass::done = false; // this is a class attribute
while (!MyClass::done) {
QApplication::processEvents();
//...
}
}
void MyClass::on_pushButton_OFF_clicked()
{
MyClass::done = true;
}
I have been trying in various ways to make my program sleep for 10 seconds before running the next line of code.
this.SetContentView (Resource_Layout.Main)
let timer = new System.Timers.Timer(10000.0)
async{do timer.Start()}
this.SetContentView (Resource_Layout.next)
I can't get any solution to work.
If you want to use async rather than the more direct way (of creating a timer and setting the content view in the event handler of the timer), then you need something like this:
this.SetContentView (Resource_Layout.Main)
async{
do! Async.Sleep(10000.0)
this.SetContentView (Resource_Layout.next) }
|> Async.StartImmediate
The key points:
Using do! Async.Sleep you can block the execution of asynchronous computation
By moving the SetContentView call inside the async, it will happen after the sleep
Using Async.StartImmediate, you start the workflow - and the sleeping ensures that the rest of the computation runs in the same threading context (meaning that it will run on the UI thread and the code will be able to access UI elements).
I have several function calls in a row that run and wait to return, then the next one runs. After these are run I have one function I want to run, but then I don't want to wait for it to be done before I run my return.
Here is an example of what I mean.
get_card, create_order, create_association and debit_order all need to wait for the previous function to complete before they can run. When I get to Queue.start_account_creation_task I want it to start running, but then let the return on the line below run right away too.
Meteor.methods({
singleDonation: function (data) {
logger.info("Started singleDonation");
//Get the card data from balanced and store it
var card = Utils.get_card(customerData._id, data.paymentInformation.href);
//Create a new order
var orders = Utils.create_order(data._id, customerData.href);
//Associate the card with the balanced customer
var associate = Utils.create_association(customerData._id, card.href, customerData.href);
//Debit the order
var debitOrder = Utils.debit_order(data.paymentInformation.total_amount, data._id, customerData._id, orders.href, card.href);
Queue.start_account_creation_task(customerData._id, data._id, debitOrder._id);
return {c: customerData._id, don: data._id, deb: debitOrder._id};
}
});
Sounds like you need parallel and serial control for tasks. The (as in, 400,000 downloads a day) Node.js module for that is called async, and a Meteor wrapper for it is peerlibrary:async.
Sooner or later you'll need a dedicated background task management package. If async is insufficient, have a look at my evaluation of packages to control background tasks in Meteor.
The thing that seemed to work the best for what I was trying to do was to just use a Meteor.setTimeout({}). It might seem like an odd choice, but it does everything I needed, including setting the Meteor Environment so that I didn't have to do any BindEnrironment call. It also breaks out of the current thread of calls, which means it then returns the result to the client and a second later finishes the rest of the calls (which are to external APIs that I didn't need my users sitting there waiting for).
I'm trying to use Dart with sqlite, with this project dart-sqlite.
But I found a problem: the API it provides is synchronous style. The code will be looked like:
// Iterating over a result set
var count = c.execute("SELECT * FROM posts LIMIT 10", callback: (row) {
print("${row.title}: ${row.body}");
});
print("Showing ${count} posts.");
With such code, I can't use Dart's future support, and the code will be blocking at sql operations.
I wonder how to change the code to asynchronous style? You can see it defines some native functions here: https://github.com/sam-mccall/dart-sqlite/blob/master/lib/sqlite.dart#L238
_prepare(db, query, statementObject) native 'PrepareStatement';
_reset(statement) native 'Reset';
_bind(statement, params) native 'Bind';
_column_info(statement) native 'ColumnInfo';
_step(statement) native 'Step';
_closeStatement(statement) native 'CloseStatement';
_new(path) native 'New';
_close(handle) native 'Close';
_version() native 'Version';
The native functions are mapped to some c++ functions here: https://github.com/sam-mccall/dart-sqlite/blob/master/src/dart_sqlite.cc
Is it possible to change to asynchronous? If possible, what shall I do?
If not possible, that I have to rewrite it, do I have to rewrite all of:
The dart file
The c++ wrapper file
The actual sqlite driver
UPDATE:
Thanks for #GregLowe's comment, Dart's Completer can convert callback style to future style, which can let me to use Dart's doSomething().then(...) instead of passing a callback function.
But after reading the source of dart-sqlite, I realized that, in the implementation of dart-sqlite, the callback is not event-based:
int execute([params = const [], bool callback(Row)]) {
_checkOpen();
_reset(_statement);
if (params.length > 0) _bind(_statement, params);
var result;
int count = 0;
var info = null;
while ((result = _step(_statement)) is! int) {
count++;
if (info == null) info = new _ResultInfo(_column_info(_statement));
if (callback != null && callback(new Row._internal(count - 1, info, result)) == true) {
result = count;
break;
}
}
// If update affected no rows, count == result == 0
return (count == 0) ? result : count;
}
Even if I use Completer, it won't increase the performance. I think I may have to rewrite the c++ code to make it event-based first.
You should be able to write a wrapper without touching the C++. Have a look at how to use the Completer class in dart:async. Basically you need to create a Completer, return Completer.future immediately, and then call Completer.complete(row) from the existing callback.
Re: update. Have you seen this article, specifically the bit about asynchronous extensions? i.e. If the C++ API is synchronous you can run it in a separate thread, and use messaging to communicate with it. This could be a way to do it.
The big problem you've got is that SQLite is an embedded database; in order to process your query and provide your results, it must do computation (and I/O) in your process. What's more, in order for its transaction handling system to work, it either needs its connection to be in the thread that created it, or for you to run in serialized mode (with a performance hit).
Because these are fairly hard constraints, your plan of switching things to an asynchronous operation mode is unlikely to go well except by using multiple threads. Since using multiple connections complicates things a lot (as you can't share some things between them, such as TEMP TABLEs) let's consider going for a single serialized connection; all activity will be serialized at the DB level, but for an application that doesn't use the DB a lot it will be OK. At the C++ level, you'd be talking about calling that execute from another thread and then sending messages back to the caller thread to indicate each row and the completion.
But you'll take a real hit when you do this; in particular, you're committing to only doing one query at a time, as the technique runs into significant problems with semantic effects when you start using two connections at once and the DB forces serialization on you with one connection.
It might be simpler to do the above by putting the synchronous-asynchronous coupling at the Dart level by managing the worker thread and inter-thread communication there. That would let you avoid having to change the C++ code significantly. I don't know Dart well enough to be able to give much advice there.
Myself, I'd just stick with synchronous connection processing so that I can make my application use multi-threaded mode more usefully. I'd be taking the hit with the semantics and giving each thread its own connection (possibly allocated lazily) so that overall speed was better, but I do come from a programming community that regards threads as relatively heavyweight resources, so make of that what you will. (Heavy threads can do things that reduce the number of locks they need that it makes no sense to try to do with light threads; it's about overhead management.)
The applications in my project were until now communicating over qtdbus using synchronous calls. However I now need to convert a few of these calls to asynchronous.
For that I chose to use this API available in qtdbus
QDBusAbstractInterface::callWithCallback
But the problem is that the current implementation has these qtdbus sync calls scattered in a lot of places in the code and the code snippets which follow these sync calls assume that the control only reaches them when the preceding call has been successfully serviced and a reply is obtained.
This will no longer be the case when the calls change to async. Moreover the calls are made in different contexts, so I will be required to maintain the state of the system before each qtdbus call, so that I know what to do when I receive the reply.
Is there any chance really to somehow convert the calls to async without rupturing the fabric of the current code in a big way?
One approach I can think of is to use the FSM pattern.
Any hints or design suggestions will be much appreciated.
Thanks!
The way I am understanding is that you will need to call the same method and then process the return value differently based on the state at the time of the call. Such as
void function()
{
//do stuff
value = SynchronousCall();
if (state == 1)
{
doSomething(value);
}
else
{
doSomethingElse(value);
}
}
I would recommend instead of a full implementation of the Finite State Machine pattern which can make a mess with the number of classes that it adds, add separate methods for each state
void function()
{
//do stuff
if (state == 1)
{
callback = *doSomething(ValueType);
}
else
{
callback = *doSomethingElse(ValueType);
}
callWithCallback(method,args, receiver,callback,error);
}
Then in each method you can assume the state and process the return value accordingly.
Another slightly (very) hacky way would be to simply have a spin wait after all the asynchronous calls and use a QThread:: yield() in the loop while you wait for the value to return. That way it is still technically an asynchronous call but it acts synchronous.