Tag Archives: reactive extensions

Batching Data by Time or Count Using Reactive Extensions (Rx)

Motivation

The time-series database InfluxDB provides a HTTP API to write data. Data points (measurements) are inserted via a Line protocol, which allows batching of data by writing multiple points in one HTTP request.

While experimenting for a simple InfluxDB C++ client, I wanted to create an asynchronous fire-and-forget API, so that the data points can be sent over HTTP without blocking the instrumented C++ code. Several “readymade” options to implement concurrency in this scenario are available.

A simple PAIR of ZeroMQ sockets would do the job, but I’d have to implement batching separately. Thus, I turned my attention to a higher-level abstraction: Rx

Rx Window Operator

Quickly looking through the cross-language Reactive Extensions site, I found the right operator: Window.

This operator has luckily been implemented in RxCpp, thus I proceeded with the experiment.

Batching Design

Batching using Rx Winodw Operator

Rx Window Operator (CC BY 3.0 reactivex.io)[1. Source: reactivex.io License: (CC BY 3.0)]

The window operator takes an observable sequence of data and splits it into windows (batches) of observables. To batch requests, the observable windows of data are aggregated to a single value upon the last value from the windows (via other aggregating Rx operators).

A Toy Problem

To validate the approach, the following problem is set:

Given a stream of integers, append the integers into a series of strings, either every second, or every N integers

String appends with integer-to-string conversions in C++ will be done via the {fmt} library.

Batching in One Line of Code

A stream of numbers batched either by time or count:

auto values = rxcpp::observable<>::range(1, 1000'000)
    .window_with_time_or_count(std::chrono::seconds(1), 100'000);

Note, there is an almost one-to-one translation into a C# version:

var values = Observable.Range(1, 1000000)
               .Window(TimeSpan.FromSeconds(1), 100000);

This indicates the power of the Rx abstraction across languages. The Rx website provides just the right sorting of the documentation to be able to translate Rx code from one language to another.

Aggregating the Batches

In order to do something useful with the batched data, the Scan operator is used to gather the data in a string buffer, and after the last value has been received, the string buffer is assembled into a string and processed:

values.subscribe(
    [](rxcpp::observable<int> window) {
        // append the number to the buffer
        window.scan(
            std::make_shared<fmt::MemoryWriter>(),
            [](std::shared_ptr<fmt::MemoryWriter> const& w, int v)
        {
            *w << v;
            return w;
        })

        // what if the window is empty? Provide at least one empty value
        .start_with(std::make_shared<fmt::MemoryWriter>())

        // take the last value
        .last()

        // print something fancy
        .subscribe([](std::shared_ptr<fmt::MemoryWriter> const& w) {
            fmt::print(
                "Len: {} ({}...)\n",
                w->size(),
                w->str().substr(0, 42)
            );
        });            
    }
);

The Tale of Two Bugs

In the initial (non-TDD) spike, the batching seemed to work, however, something caught my attention (the code bites back):

[window 0] Create window
Count in window: 170306
Len: 910731 (123456789101112131415161718192021222324252...)

the window wasn’t capped at 100’000. This could have been either a misunderstanding or a bug, thus I formulated a hesitant issue #277. As it turned out, it indeed was a bug, which was then fixed in no time. However, the first bug has hidden another one: the spike implementation started to crash at the end: when all the windows were capped by count, and not by time, last window was empty, as all values fit exactly into 10 batches.

The Last operator rightly caused an exception due to an empty sequence. Obviously, there’s no last value in an empty sequence. Rubber Ducking and a hint from Kirk Shoop fixed the issue by utilizing the StartWith operator to guarantee, the sequence is never empty. An empty string buffer can be ignored easier downstream.

Active Object

The active object pattern was applied to implement a fire-and-forget asynchronous API. A Rx Subject to bridge between the function call and the “control-inverted” observable:

struct async_api {
    //...
    rxcpp::subjects::subject<line> subj;
    //...

    async_api(...)
    {
        auto incoming_requests = subj
            .get_observable()
            .map([](auto line) {
                return line.to_string();
            });

        incoming_requests
            .window_with_time_or_count(
                window_max_ms,
                window_max_lines,
                // schedule window caps on a new thread
                rxcpp::synchronize_new_thread()
            )
            .subscribe(...)
        ;
    }

    // fire-and-forget
    void insert(line const& line)
    {
        subj
            .get_subscriber()
            .on_next(line);
    }
};

in order not to block the caller (which would be the default behavior), the observable watches the values from each window on a new thread. Here, scheduling on a thread pool (currently missing in RxCpp) would probably be beneficial.

While this implementation might not be an optimal one, the declarative nature of Rx, once the basics are understood, allows to “make it work and make it right” pretty quickly by composing the right operators.

Code

The runnable code of the example can be found at Github: C++ version.

In order to show, how similar the high level code can be between different languages when writing, I’ve “ported” the example to C# [2. The C# version appears to run faster on my windows machine while solving the same toy problem].

Why wait forever for the tests? Fast tests of slow software.

Time is volatile

Imagine writing a cron-like functionality that should produce some side-effect, such as cleanup. The intervals between such actions might be quite long. How does one test that? One can surely reason about the software, but given a certain complexity, test should be written, proving that certain important scenarios work as intended.

It’s common that software depends on time flow as dictated by the physical time flow, reflected via some clock provider. However, resetting the time to a year ahead won’t make the CPU work faster and make all the computations it should have performed within that year. A clock is also a volatile component that can be manipulated, thus if time is an issue, it’s probably a good idea not to depend on it directly, following the Stable Dependencies Principle and the Dependency Inversion Principle.

Luckily, there is an abstraction for time, at least in Reactive Extensions (Rx), which is the Scheduler.

Slow non-tests

Here’s a slow Groovy non-test, waiting for some output on the console using RxGroovy:

import rx.*
import java.util.concurrent.TimeUnit

def observable = Observable
	.just(1)
	.delay(5, TimeUnit.SECONDS)

observable.subscribe { println 'ah, OK, done! Or not?' }

Observable
	.interval(1,TimeUnit.SECONDS)
	.subscribe { println 'still waiting...' }

println 'starting to wait for the test to complete ...'

observable.toBlocking().last()

Running it produces the following slow-ticking output:

oldnontest [1. Caputured with the wonderful pragmatic tool LICEcap by the Reaper developers]

Interpreting such tests without color can be somewhat challenging [2. Here, the ‘still waiting’ subscription is terminated after the first subscription ends. Try exchanging the order of the subscribe calls.].

Fast tests

Now let’s test something ridiculous, such as waiting for a hundred days using Spock. Luckily, RxJava & RxGroovy also do implement the test scheduler, thus enabling fast tests using virtual time:

import spock.lang.Specification

import rx.Observable
import rx.schedulers.TestScheduler
import java.util.concurrent.TimeUnit


class DontWaitForever extends Specification {
    def "why wait?"() {
        setup:
            def scheduler = new TestScheduler()

            // system under test: will tick once after a hundred days
            def observable = Observable
                .just(1)
                .delay(100, TimeUnit.DAYS, scheduler)
            def done = false

        when:
            observable.subscribe {
                done = true
            }

            // still in the initial state
            done == false

        and:
            scheduler.advanceTimeBy 100, TimeUnit.DAYS

        then:
            done == true
    }
}

fasttest [3. Building using Gradle]

just checking, advancing the time by 99 days results in a failure:

just_checking

Delightful, groovy colors!

Source

github.com/d-led/dont_wait_forever_for_the_tests

Deterministic Testing of Concurrent Behavior in RxCpp

A Retrospective

After getting inspired by The Reactive Manifesto, it is hard not to get excited about Reactive Extensions. Such excitement has lead to a series of hello-world articles and some code examples. While Reactive Extensions take over the programming world in C#, Java and JavaScript, it seems, the world of C++ is slow to adopt RxCpp.

The new ReactiveX Tutorial link list is a great place to start learning and grokking. This article is an attempt to bring RxCpp closer to C++ developers who might not see yet, how a reactive programming model might help writing better, more robust code.

Testing concurrency with RxCpp

A previous article showed how to test ViewModels in C# by parameterizing the ViewModels with a scheduler. In a UI setting, the scheduler usually involves some kind of synchronization with the GUI thread. Testing keystrokes arriving at certain speed would require some effort to simulate events, probably leading to brittle tests. With the scheduler abstraction, the concurrent behavior of a component is decoupled from physical time, and thus can be tested repeatedly and very fast. This was the C# test:

(new TestScheduler()).With(scheduler =>
{
    var ticker = new BackgroundTicker(scheduler);

    int count = 0;
    ticker.Ticker.Subscribe(_ => count++);
    count.Should().Be(0);

    // full control of the time without waiting for 1 second
    scheduler.AdvanceByMs(1000);
    count.Should().Be(1);
});

Show Me The Code

Without further ado, the C++ version is not very far from the C# version. In a simple test, we can parameterize a sequence of integer values arriving at specified intervals (a ticker) with a coordination (why coordination and not scheduler, read in the RxCpp developer manual:

auto seq = rxcpp::observable<>::interval(
            std::chrono::milliseconds(1),
            some_scheduler
);

The deterministic test scheduler API is currently available through a worker created on the test scheduler:

auto sc = rxcpp::schedulers::make_test();
auto worker = sc.create_worker();
auto test = rxcpp::identity_same_worker(worker);

The rest should read like English:

int count = 0;

WHEN("one subscribes to an observable sequence on the scheduler") {
  auto seq = rxcpp::observable<>::interval(
              std::chrono::milliseconds(1),
              test // on the test scheduler
             ).filter([](int i) { return i % 2; });

  seq.subscribe([&count](int){
    count++;
  });

  THEN("the sequence is not run at first") {
    worker.sleep(2 /* ms */);

    CHECK(count == 0);

    AND_WHEN("the test scheduler is advanced manually") {

      THEN("the sequence is run as expected") {
        worker.advance_by(8 /* ms */);
        CHECK(count == 5);
      }
    }
  }
}

The full test can be seen @github, and is built on Travis CI