<p>Recently ReactJS v18 was released. I started to look into its improvements, and checked which of my projects could benefit from them.</p><p>The main react-based project I maintain at the moment of writing this article is <a target="_blank" rel="noopener noreferrer" href="https://github.com/shlinkio/shlink-web-client">shlink-web-client</a>, so I naturally created a new branch and started the process.</p><p>The update was quite smooth, as most of the other dependencies I was using from the react ecosystem were already updated to support v18. All but one, <a target="_blank" rel="noopener noreferrer" href="https://enzymejs.github.io/enzyme/">enzyme</a>, the testing framework I was using to unit-test react components.</p><p>To be precise, I was able to continue running my tests, but some of them (those using <code>mount</code> instead of <code>shallow</code>) were still rendering components with the &quot;old&quot; React v17 approach, causing them to behave as in that version and a warning being printed in the console.</p><h3>First fix attempt</h3><p>The way enzyme integrates with the react version you are using is through an adapter library. The last official one supports only React v16, but someone released an <a target="_blank" rel="noopener noreferrer" href="https://github.com/wojtekmaj/enzyme-adapter-react-17">unofficial adapter for v17</a> which, with close to 650k weekly downloads, quickly became the way to go.</p><p>I checked if there was one for React 18, but sadly there wasn&#x27;t.</p><p>Also, the author of the adapter for v17 posted an article indicating <a target="_blank" rel="noopener noreferrer" href="https://dev.to/wojtekmaj/enzyme-is-dead-now-what-ekl">enzyme is dead</a>, he was not going to release one for v18, and publishing one for v17 was a mistake.</p><p>With those facts in mind, I started to face what was probably going to be the moment to migrate to <a target="_blank" rel="noopener noreferrer" href="https://testing-library.com/docs/react-testing-library/intro">react testing library</a>.</p><h3>Migration strategy</h3><p>I decided to follow a migration strategy similar to the one suggested in the article above, but with a couple of modifications:</p><ul><li>I update to React 18 anyway, as tests still pass.</li><li>Migrate right away all tests using <code>mount</code>, as those are the ones behaving differently.</li><li>New tests should be written directly with react testing library (RTL from now on).</li><li>When an existing test needs to be changed, consider migrating it to RTL, and do it if there&#x27;s time.</li><li>Introduce some PR from time to time just migrating a handful of tests.</li><li>Once everything is migrated, get rid of everything related with enzyme.</li></ul><h3>First contact and impressions</h3><p>I was a bit sceptical with RTL, as it was focused more on integration tests, instead of unit tests, which meant I had to completely change the mindset to write tests.</p><p>It also didn&#x27;t support shallow rendering, meaning that you will end up sometimes rendering a bigger component tree when testing container-like components, potentially causing side effects that could affect your test.</p><p>On the other hand there&#x27;s more and more people stating that <a target="_blank" rel="noopener noreferrer" href="https://kentcdodds.com/blog/why-i-never-use-shallow-rendering">shallow rendering is an anti-pattern</a>, and that&#x27;s probably the reason of enzyme getting &quot;discontinued&quot;.</p><p>Also, RTL has become the <a target="_blank" rel="noopener noreferrer" href="https://reactjs.org/docs/testing.html#tools">officially recommended framework</a> to test react components, it has become very popular, and it has even joined a wider group of testing libraries under the <a target="_blank" rel="noopener noreferrer" href="https://testing-library.com/"><code>@testing-library</code></a> organization.</p><p>The thing is that I started to use it, and I was quickly surprised with how it felt to work with it, and all the benefits it was bringing to my tests.</p><h3>Benefits I have experienced</h3><p>Once I started migrating the first tests I noticed some benefits:</p><ul><li>Changing my mindset was actually easier than I thought. The docs are very well written, so it&#x27;s relatively easy to start using RTL.</li><li>My tests got simpler, and required less of those nasty global mocks that jest allows (I&#x27;m looking at you, <a target="_blank" rel="noopener noreferrer" href="https://jestjs.io/es-ES/docs/mock-functions#mocking-modules"><code>jest.mock</code></a>).</li><li>Tests were way less coupled with implementation details. I moved from testing the component props, to test what the user would see on the screen.</li><li>Related with previous point, I started to write my tests by looking at the screen instead of the component&#x27;s implementation.</li><li>Changes in source code led to fewer tests requiring to be changed.</li><li>A progressive migration was possible, where some tests still use enzyme and others are already using RTL.</li></ul><p>And overall, it felt good and I have already dedicated a couple of PRs just to migrate more tests.</p><h3>Challenges I faced</h3><p>But of course, migrating to a new tool always comes with its own challenges, and this was not an exception. I&#x27;ll explain the main ones I have faced so far and how I tackled them:</p><ul><li><p>RTL lets you do some things in a couple of different ways, and the docs may not be super clear about what&#x27;s the recommended option.</p><p>For this I recommend reading <a target="_blank" rel="noopener noreferrer" href="https://acel.me/DCMb1">this article</a> from Kent C. Dodds, RTL&#x27;s author, which explains some common mistakes when using the library.</p></li><li><p>At first, I struggled a bit with asynchronous side effects and state changes in components, which can easily end up printing a warning like <code>Warning: An update to MyComponent inside a test was not wrapped in act</code>.</p><p>The warning also explains how you are supposed to solve this by wrapping your code in <code>act()</code>, but this warning comes from React, not RTL, and what it is not telling you is that RTL already wraps all needed operations in <code>act()</code>.</p><p>What this error usually mean is that you are not waiting for something to happen. <a target="_blank" rel="noopener noreferrer" href="https://acel.me/jnUdT">This other article</a>, also from Kent C. Dodds, explains everything you need to know about the things that can lead to this warning, and how to solve them.</p></li><li><p>Since you start testing from the user point of view, some use cases can be challenging to test without coupling with implementation details. For example, when you have a component which dynamically renders different images or icons, I used to directly check the component properties.</p><p>For this I found the best solution was to use <a target="_blank" rel="noopener noreferrer" href="https://jestjs.io/docs/snapshot-testing">jest snapshots</a>. I didn&#x27;t want to manually check which SVG was being rendered in the DOM, just that different ones were being used in every case.</p><pre><pre><code class="language-tsx">// I went from this...
test(&#x27;...&#x27;, () =&gt; {
  const wrapper = shallow(&lt;MyComp dynamicValue=&quot;foo&quot; /&gt;);
  expect(wrapper.find(FontAwesomeIcon).prop(&#x27;icon&#x27;)).toEqual(someIcon);
})

// To this
test(&#x27;...&#x27;, () =&gt; {
  const { container } = render(&lt;MyComp dynamicValue=&quot;foo&quot; /&gt;);
  expect(container.firstChild).toMatchSnapshot();
})
</code></pre></pre></li><li><p>Something very similar happens if you use some library that renders canvas. There&#x27;s no way to test that from a DOM point of view, which is what RTL does.</p><p>In my case I was using <a target="_blank" rel="noopener noreferrer" href="https://www.chartjs.org/">Chart.js</a> to render some charts, so I used <a target="_blank" rel="noopener noreferrer" href="https://github.com/hustcc/jest-canvas-mock">jest-canvas-mock</a>, which exposes a method on the canvas to see which are all the events invoked from the library.</p><p>Then I used snapshots again to verify they had the expected &quot;shape&quot;.</p><pre><pre><code class="language-tsx">// I went from this...
test(&#x27;...&#x27;, () =&gt; {
  const wrapper = shallow(&lt;MyCompWithCanvas /&gt;);
  expect(wrapper.find(Chart).prop(&#x27;data&#x27;).datasets).toEqual(...);
})

// To this
test(&#x27;...&#x27;, () =&gt; {
  const { container } = render(&lt;MyCompWithCanvas /&gt;);
  const events = container.querySelector(&#x27;canvas&#x27;)?.getContext(&#x27;2d&#x27;)?.__getEvents();

  expect(events).toBeTruthy();
  expect(events).toMatchSnapshot();
})
</code></pre></pre></li></ul><h3>Conclusion</h3><p>For better or worst, enzyme should no longer be considered an option. If you are starting a new project or are using it on small ones, consider migrating to RTL as soon as possible.</p><p>If you are using it in bigger projects, following the process described in this article will probably help you.</p><p>In any case, you will get surprised. React testing library is a great tool and provides many benefits over enzyme.</p><p>Its main issue is that it&#x27;s not a drop-in replacement for enzyme, so you will have to change a bit your mindset.</p>

My experience migrating from enzyme to react testing library

<p>Capturing code coverage for a test suite is a very useful way to know which parts of your source code are actually getting executed by tests.</p><p>This is useful not only to know if you need to add more tests to your project (in case the coverage is too low), but also, if you want to apply technics like mutation testing, having a properly generated code coverage report will dramatically speed up te execution and help you know if a non-killed mutant is relevant.</p><h3>Generating the code coverage</h3><p>Capturing code coverage is usually easy in the most common types of tests, like unit tests, where the source code is executed in the same process.</p><p>Tools like <a target="_blank" rel="noopener noreferrer" href="https://phpunit.de/">PHPUnit</a>, one of the most extended PHP testing frameworks, ships with tools like <a target="_blank" rel="noopener noreferrer" href="https://github.com/sebastianbergmann/php-code-coverage">php-code-coverage</a>, designed to capture the code under execution, just by providing a couple of lines of configuration.</p><p>However, generating a reliable code coverage report is not always that easy for other types of tests.</p><p>In contexts like E2E tests, where you may spin-up an API on a different process, and make your tests just call this API via HTTP and assert on the responses, PHPUnit will not be able to capture the code coverage, because the source code is actually being loaded on a different process.</p><h3>Remote code coverage</h3><p>A couple of weeks ago, I faced the issue described above. I had a web service for which I had a decent API test suite, but I wanted to capture code coverage for two main reasons:</p><ul><li>Be able to merge this coverage with the one generated by other tests suites, to know what&#x27;s the &quot;total&quot; code coverage, regardless the test producing it.</li><li>Apply mutation testing to the API tests, as I was already doing that for the rest of the test suites in the project. For this, I also needed to know exactly what tests were covering every line of the source code.</li></ul><p>After some investigation, I couldn&#x27;t find a built-in feature, where PHPUnit itself was able to &quot;remotely&quot; capture te code under test, so I had to think on a way to manually use <code>php-code-coverage</code> by myself.</p><h3>Code coverage reports</h3><p>In order to be able to capture code coverage, I had to make sure the instances of the objects provided by <code>php-code-coverage</code> where &quot;spin-up&quot; with the server, and then, the code coverage report was dumped at the end.</p><p>Also, you usually define the code to include/exclude from coverage in PHPUnit&#x27;s config, but since this time we are the ones creating the coverage objects, this was not possible.</p><p>I tried a couple of different approaches (more on this at the end), but the most accurate one for me, was to define two routes in my server, that were dynamically configured only when running the server in the context of these tests, and were responsible for creating and dumping the code coverage.</p><pre><pre><code class="language-php">use PHPUnit\Runner\Version;
use SebastianBergmann\CodeCoverage\CodeCoverage;
use SebastianBergmann\CodeCoverage\Driver\Selector;
use SebastianBergmann\CodeCoverage\Filter;
use SebastianBergmann\CodeCoverage\Report\Html\Facade as Html;
use SebastianBergmann\CodeCoverage\Report\PHP;
use SebastianBergmann\CodeCoverage\Report\Xml\Facade as Xml;

// Define directories containing the code you want to cover
$filter = new Filter();
$filter-&gt;includeDirectory(...);
$filter-&gt;includeDirectory(...);
$coverage = new CodeCoverage((new Selector())-&gt;forLineCoverage($filter), $filter);

// This is &quot;pseudo-code&quot;. Defining the routes depends on the framework you use
$app-&gt;get(&#x27;/api-tests/start-coverage&#x27;, function () use (&amp;$coverage) {
    $coverage-&gt;start(&#x27;API tests&#x27;);
    return new EmptyResponse();
});
$app-&gt;get(&#x27;/api-tests/stop-coverage&#x27;, function () use (&amp;$coverage) {
    $basePath = &#x27;...&#x27;; // Wherever you want the reports to be dumped
    $coverage-&gt;stop();

    // I generated coverage reports in a couple of different formats
    (new PHP())-&gt;process($coverage, $basePath . &#x27;/coverage.cov&#x27;);
    (new Xml(Version::getVersionString()))-&gt;process($coverage, $basePath . &#x27;/coverage-xml&#x27;);
    (new Html())-&gt;process($coverage, $basePath . &#x27;/coverage-html&#x27;);

    return new EmptyResponse();
});
</code></pre></pre><p>Then, I added the next snippet to my PHPUnit <code>bootstrap.php</code> file, to make sure the endpoints configured above are invoked at the proper times:</p><pre><pre><code class="language-php">// when starting up the test suite, call the endpoint to start collecting coverage
$httpClient-&gt;request(&#x27;GET&#x27;, &#x27;http://localhost:8080/api-tests/start-coverage&#x27;);

// When the tests process shuts down, call the endpoint to stop the coverage
register_shutdown_function(function () use ($httpClient): void {
    $httpClient-&gt;request(&#x27;GET&#x27;, &#x27;http://localhost:8080/api-tests/stop-coverage&#x27;);
});
</code></pre></pre><blockquote><p>Learn more about how to configure PHPUnit&#x27;s <a target="_blank" rel="noopener noreferrer" href="https://phpunit.readthedocs.io/en/9.5/configuration.html#the-bootstrap-attribute">bootstrap script</a>.</p></blockquote><p>This was fine from a pure code coverage point of view, but if you try it yourself, you will notice the report says all the code has been covered by <code>API tests</code>, which is what we provide when calling <code>$coverage-&gt;start(&#x27;API tests&#x27;);</code></p><p><img src="/assets/img/remote-coverage/invalid-api-coverage.png" alt="Invalid API coverage"/></p><p>To properly do mutation testing we need to know which are the exact tests that cover every part of the code. That requirement took me to the next approach.</p><h3>Supporting mutation testing</h3><p>To be more precise with the generated reports, I needed some way to notify the server about the specific test performing the request.</p><p>One way to do this is by sending a custom header when performing a request from your tests, and registering some event handler or middleware which starts the coverage for that request with the value coming from that header.</p><p>With that, I evolved the configuration above, and added a middleware handler (the framework I use is <a target="_blank" rel="noopener noreferrer" href="https://www.php-fig.org/psr/psr-15/">PSR-15</a> based, but the framework of your choice will probably have a way to do this).</p><pre><pre><code class="language-php">use ...

$coverage = new CodeCoverage();

// This ensures the coverage is started and stopped on every request, using the &quot;x-coverage-id&quot; header as the ID.
$app-&gt;middleware(function (ServerRequestInterface $req, RequestHandlerInterface $handler) use (&amp;$coverage) {
$coverage?-&gt;start(&#x27;API Tests&#x27;);
    $coverage-&gt;start($req-&gt;getHeaderLine(&#x27;x-coverage-id&#x27;));

    try {
        return $handler-&gt;handle($req);
    } finally {
        $coverage-&gt;stop();
    }
});

// This route is no longer needed. You can also remove the request from your bootstrap script
// $app-&gt;get(&#x27;/api-tests/start-coverage&#x27;, ...);
$app-&gt;get(&#x27;/api-tests/stop-coverage&#x27;, ...);
</code></pre></pre><p>Then, inside your test you can do something like this:</p><pre><pre><code class="language-php">use PHPUnit\Framework\TestCase;

class MyApiTest extends TestCase
{
    private ClientInterface $httpClient;

    public function setUp(): void
    {
        $this-&gt;httpClient = ...;
    }

    public function testMyEndpoint(): void
    {
        $resp = $this-&gt;httpClient-&gt;request(&#x27;POST&#x27;, &#x27;...&#x27;, [
            &#x27;headers&#x27; =&gt; [
                &#x27;x-coverage-id&#x27; =&gt; static::class,
            ],
        ])

        // Asserts go here...
    }
}
</code></pre></pre><p>After this, the code coverage reports will properly show the actual tests covering them.</p><p><img src="/assets/img/remote-coverage/valid-api-coverage.png" alt="Valid API coverage"/></p><h3>Alternative approach</h3><p>The approach explained here is functional, but it&#x27;s a bit hacky. I had to do it like this because of limitations with how I serve the API with <a target="_blank" rel="noopener noreferrer" href="https://openswoole.com">openswoole</a>.</p><p>However, instead of defining routes that need to be actively called to start/stop the code coverage (or just to dump the report, as in the second approach), you could try to register the shutdown function directly in the server side.</p><p>If that works for you, it&#x27;s much more straightforward, and you won&#x27;t need to bother changing your tests bootstrap script.</p><h3>Conclusion</h3><p>Even though there&#x27;s room for improvement, this allows to generate code coverage reports for code that&#x27;s running on a different process.</p><p>I plan to apply something similar for E2E tests for a command line app, so maybe I extend this article at some point.</p><p>Also, I decided to write this article, because at the moment of having to implement this myself, I didn&#x27;t find too much useful information out there.</p><p>I hope this helped you, and now you can collect code coverage reports on your API tests.</p>

Capturing remote code coverage in API tests with PHPUnit

<p>In the last couple of days I have been playing around with the idea of combining GitHub Actions and GitHub Pages to be able to dynamically build preview environments every time a pull request is opened against a repository hosting a client-side web app.</p><p>The idea is that you can quickly preview the changes made by some contributor, without having to clone their branch and running everything locally, something specially useful for open source projects.</p><p>In this article I will explain how to set everything up, and what are some considerations and problems you may face during the process.</p><h3>Enable github pages</h3><p>The first thing you need to do is enable your project to publish pages on some branch. I recommend creating a new branch for this, let&#x27;s say <code>preview-env</code>, and then enabling publishing on it.</p><p>In order to do that go to your repository and then <strong>Settings</strong> -&gt; <strong>Pages</strong>, look for the <strong>Source</strong> section, and you should see something like this:</p><p><img src="/assets/img/preview-envs/github-pages-before-enabling.png" alt="GitHub pages before enabling"/></p><p>Click in the dropdown and select your new branch, <code>preview-env</code>. Then click <strong>Save</strong>.</p><p><img src="/assets/img/preview-envs/github-pages-after-enabling.png" alt="GitHub pages after enabling"/></p><p>After saving you will see a message at the top of this section telling you what&#x27;s the URL in which the pages are published. It always follows the same pattern. If your organization is <code>my-org</code> and your repository is <code>my-repo</code>, the url will be <code>https://my-org.github.io/my-repo/</code>.</p><h3>Configure the workflow</h3><p>Now we need to configure a workflow that will &quot;deploy&quot; the preview environment for the branch every time a new pull request is created or updated:</p><p>Let&#x27;s imagine we have a React web app that is built with node.js. The configuration could look like this.</p><pre><pre><code class="language-yml"># .github/workflows/deploy-preview.yml
name: Deploy preview

on:
  pull_request: null

jobs:
  deploy:
    runs-on: ubuntu-20.04
    continue-on-error: true
    steps:
      - name: Checkout code
        uses: actions/checkout@v2
      - name: Use node.js 14.15
        uses: actions/setup-node@v1
        with:
          node-version: 14.15
      - name: Generate slug
        id: generate_slug
        run: echo &quot;##[set-output name=slug;]$(echo ${GITHUB_HEAD_REF#refs/heads/} | sed -r &#x27;s/[~\^]+//g&#x27; | sed -r &#x27;s/[^a-zA-Z0-9]+/-/g&#x27; | sed -r &#x27;s/^-+\|-+$//g&#x27; | tr A-Z a-z)&quot;
      - name: Build
        run: npm ci &amp;&amp; npm run build
      - name: Deploy
        uses: JamesIves/github-pages-deploy-action@4.1.1
        with:
          branch: preview-env
          folder: build
          target-folder: ${{ steps.generate_slug.outputs.slug }}
      - name: Publish env
        uses: marocchino/sticky-pull-request-comment@v2
        with:
          header: Preview environment
          message: |
            ## Preview environment
            https://my-org.github.io/my-repo/${{ steps.generate_slug.outputs.slug }}/
</code></pre></pre><blockquote><p>If you are not familiar with GitHub Actions yet, take a look at the <a target="_blank" rel="noopener noreferrer" href="https://docs.github.com/en/actions">official documentation</a>.</p></blockquote><p>This workflow will follow the next steps:</p><ul><li>Checkout the code from the repository.</li><li>Set-up node.js 14.15</li><li>Based on the branch name, generate a slug that we will use to have separated deployment folders for every pull request we receive. The slug is exposed to next steps as an output.</li><li>Build the project. This step will obviously depend on the needs of your project.</li><li>Using <a target="_blank" rel="noopener noreferrer" href="https://github.com/JamesIves/github-pages-deploy-action">JamesIves/github-pages-deploy-action</a>, deploy the contents we previously built in the GitHub page, under a sub-folder matching the slug we generated a few steps above, and under the branch <code>preview-env</code> (remember we enabled GitHub pages for this branch).</li><li>Finally, using <a target="_blank" rel="noopener noreferrer" href="https://github.com/marocchino/sticky-pull-request-comment">marocchino/sticky-pull-request-comment</a>, post a comment in the pull request with a link to the preview environment, so that you can just click on it and see the deployment.</li></ul><p>The steps may change depending on your needs and how you build your project. For example, I have an extra step to update a property in the <code>package.json</code> that is then used to allow serving the app in a sub-path instead of the root of the domain.</p><p>You may also not build your site with node.js, as you could be using any of the multiple existing static site generators out there, but you get the idea.</p><p>Finally, you have probably noticed the config includes <code>continue-on-error: true</code>. This will allow the pull request to be merged even if this process fails for some reason.</p><h3>Considerations when using forks</h3><p>While the configuration above will work perfectly when creating pull requests inside the same repository, there are some limitations when the pull request comes from a fork (which is the usual in OSS projects).</p><p>For security reasons, the <code>GITHUB_TOKEN</code> injected in those cases doesn&#x27;t have write permissions, which will make the &quot;deploy&quot; step fail.</p><p>Also, you cannot define your own token with more permissions and inject it as a secret, as the secrets are not exposed to workflows running on forks.</p><p>However, with a couple small config changes, we can work around these limitations:</p><h4>Change event</h4><p>The first thing we need to do is use the <code>pull_request_target</code> event instead of the <code>pull_request</code> one.</p><pre><pre><code class="language-diff">on:
-  pull_request: null
+  pull_request_target: null
</code></pre></pre><p>According to the documentation, this is equivalent to the <code>pull_request</code> event, but it runs in the scope of the base repository.</p><p>Thanks to this, the <code>GITHUB_TOKEN</code> will come with write permissions, and we will have access to secrets if needed.</p><h4>Checkout the fork</h4><p>The second thing we are going to change is that we now need to specify the repository to checkout.</p><p>In order to do that, we need to add two inputs to the &quot;Checkout code&quot; step, like this:</p><pre><pre><code class="language-diff">    steps:
      - name: Checkout code
        uses: actions/checkout@v2
+       with:
+         repository: ${{ github.event.pull_request.head.repo.full_name }}
+         ref: ${{ github.event.pull_request.head.ref }}
</code></pre></pre><p>This will ensure we checkout from the fork repository, instead of the base one, which is the default, as it&#x27;s the one where the <code>pull_request_target</code> event is scoped to.</p><p>And that&#x27;s it. With these two changes you can now deploy preview environments for pull requests coming from forks.</p><blockquote><p><strong>IMPORTANT!</strong> The limitations of the <code>pull_request</code> event are there for a reason. People could potentially get access to your secrets or manipulate your code. Proceed at your own risk.</p></blockquote><h3>Other considerations</h3><p>Other than the limitations for forks, there are a few other minor things to take into consideration.</p><ul><li>Changes in GitHub Pages do not reflect immediately. It may take up to a couple of minutes for the deployment to be available, but considering you will usually not review a PR right after it has been created, this should not be a big issue.</li><li>As you have seen, every time a pull request is created, the deployment will generate a new folder in your GitHub Pages branch. This may result in a lot of old folders to pile-up, so you may want to set up some way to clean them up periodically or after the pull request has been merged.</li><li>Due to the fact that GitHub Pages only allows static pages to be served, this process cannot be used with apps that require some kind of back-end or server-side logic.</li><li>If your project is already using GitHub Pages for something else (serve the project&#x27;s website or docs), that may conflict with this, preventing you from having this kind of preview envs.</li></ul>

Setting-up a preview environment for your client-side apps with GitHub Actions and GitHub Pages

<p>I remember many years ago, when I released my first open source project, that every time I had a new release ready, I was never sure what should be the new version number.</p><p>Versioning software is not as straightforward as it might look when you don&#x27;t have a lot of experience.</p><p>In this article I&#x27;ll try to explain the <a target="_blank" rel="noopener noreferrer" href="https://semver.org/">SemVer</a> approach to version software, and some things you might want to take into account on top of that.</p><h3>Semantic Versioning</h3><p>The first thing you need is a solid set of rules that help you decide which should your next version be.</p><p>I remember thinking <em>&quot;is this version big enough as to jump from 1.0 to 2.0?&quot;</em> or <em>&quot;I have seen projects using three numbers for the version, X.Y.Z, should I do the same?&quot;</em></p><p>Those questions are easier to answer if you stick to some standard that tells you what to do, and that will help others more easily understand your intention.</p><p>My recommendation is to use <a target="_blank" rel="noopener noreferrer" href="https://semver.org/">Semantic Versioning</a>, as it defines a solid starting point, and it&#x27;s quite adopted by the community.</p><p>Semantic Versioning states that you should use three numbers to identify versions, <code>MAJOR.MINOR.PATCH</code>, and follow these rules in order to decide which one to increase:</p><ul><li><code>PATCH</code>: the new release includes only bug fixes. People are encouraged to update to this release and no changes are needed in their code bases.</li><li><code>MINOR</code>: the new release includes new backwards-compatible features. People can update to this version without changing their code bases, and make use of the new features afterwards if they wish.</li><li><code>MAJOR</code>: the new release includes <a target="_blank" rel="noopener noreferrer" href="https://en.wikipedia.org/wiki/Backward_compatibility">backward compatibility</a> breaks. Changes in the code base will be needed when updating to this version.</li></ul><blockquote><p>These rules are easier to apply to libraries and frameworks than it is for apps and services. More about this later.</p></blockquote><h3>Try to keep backwards-compatibility</h3><p>Ok, we have some rules now. However, I have seen very frequently people mistaking what <em>can</em> be done with what <em>should</em> be done.</p><p>Yes, if you introduce a breaking change, you can update the major version and that&#x27;s it. However, if you do it too frequently, people won&#x27;t be happy, and you will end up with a fragmented user base, that stop updating to newer versions.</p><p>I remember using a library to deploy stuff that had a new major version every couple of months, changing things like &quot;The <code>host</code> function is now called <code>server</code>&quot;. Yes, ok, but why do I have to change my stuff so frequently.</p><p>It&#x27;s usually better to try to implement things in a backwards-compatible way, and keep track of the things you want to change and do it all at a time.</p><p>You can also deprecate things, and document the way you want those things to be done moving forward. That way you give people the chance to start changing their code and simplify a hypothetical upcoming major release.</p><h3>Provide clear upgrade paths</h3><p>As mentioned above, too frequent major version changes are not desirable, but sooner or later you will want to release a new major version for different reasons.</p><ul><li>Getting rid of that feature which is hard to maintain, introduces lots of bugs, and provides little value.</li><li>Supporting something new which is very hard to implement without breaking backwards compatibility.</li><li>Removing all those deprecated methods you have been dragging for months.</li><li>Fix design issues, where something was architected in the wrong way, and only spotted after some time.</li></ul><p>At the time this happens you will want your users to upgrade to the newer version, as you don&#x27;t want people to report bugs that only affect older versions, or to expect you to backport fixes to older versions because they haven&#x27;t updated to the new one.</p><p>The best way to achieve this is to provide clear and simple upgrade paths:</p><ul><li><p>Prepare your releases for the things to come: If you want to introduce new ways of doing things, do it before the major release, and deprecate the older ones. For example, you can have a new method that gets called by an older one, while you mark the old one as deprecated.</p><p>People will be more willing to update and change their code afterwards. By the time the major release is published, their code will be ready to upgrade.</p></li><li><p>Include a CHANGELOG file in your project, explaining the changes introduced in every version. A very good way to document a change log is the <a target="_blank" rel="noopener noreferrer" href="https://keepachangelog.com/">Keep a changelog</a> initiative.</p><p>You can also use some other platform to document the change log, like GitHub releases.</p></li><li><p>Include documentation on how to upgrade to a major release: This is probably the most important one. Document all the steps that need to be followed and all the required changes to update from a release to the next major one.</p><p>A good way to do this is by including an UPGRADE file in your project. You can see <a target="_blank" rel="noopener noreferrer" href="https://github.com/acmailer/acmailer/blob/main/UPGRADE.md">this example</a>.</p></li></ul><h3>Versioning libraries vs apps</h3><p>Semantic Versioning is a good standard to version software, but it plays much better with libraries and frameworks that are installed using some kind of package manager, than it does with apps and services that are not directly depended on.</p><p>For example, how do you identify a backward compatibility break on a web user interface?</p><p>With this kind of software there are more grey areas, and you will end up using your intuition.</p><p>For example, not long ago I released a <a href="/2020/12/24/released-shlink-web-client-v3-0-0/">new major version</a> for a web app I maintain, and it was completely compatible with the previous one. However, it introduced a larger amount of features than usual, and big UI changes, so I decided it was worth the bump.</p><p>Because of this there are projects that use different approaches to version apps. For example:</p><ul><li>Firefox and Chrome have 6-week cycles, and after each one of them they release a new major version no matter what.</li><li>Jetbrains publishes a few releases of their IDEs every year (I think 3) which always start with the year they were released on, like <code>2020.3.0</code>, and they increase the patch number for bugfixes.</li><li>Ubuntu releases 2 versions every year, one in April and one in October, and they are always identified by the year and month they were released on, like <code>19.04</code> or <code>20.10</code>.</li></ul><h3>Special versions</h3><p>There&#x27;s also a set of special versions that can be used to publish the so-called <strong>pre-release</strong> versions. These <a target="_blank" rel="noopener noreferrer" href="https://semver.org/#spec-item-9">are covered</a> by Semantic Versioning, but sometimes they are not so easy to understand.</p><p>It is also very likely that you never have to use these, as only relatively big projects will need this level of granularity, but it&#x27;s still good to know what&#x27;s their purpose and what do they imply.</p><h4>0.x.y</h4><p>It is possible to release a version of your software where the major number is <code>0</code>. When this happens, it means your software is still in its early stages.</p><p>It does not mean it&#x27;s necessarily unstable, but it means you are still introducing frequent changes on it, and therefore, it might be a bit more fragile than a version starting with a number greater than <code>0</code>.</p><h4>Alpha</h4><p>The intention of an alpha release is that a subset of people con test some new functionality, but you don&#x27;t want people to use it in production</p><p>Alpha versions are on their very early stages. Their public API is very likely to change, and they probably contain bugs.</p><p>Semantic Versioning states that alpha versions are identified by this pattern: <code>MAJOR.MINOR.PATCH-alpha.NUMBER</code>, for example <code>2.0.0-alpha.1</code>.</p><h4>Beta</h4><p>Similar to alpha versions, beta versions are also prone to contain bugs, but their API will in theory not change. That means that you could potentially adapt your software to start using a beta version, and when the corresponding stable release is published, you shouldn&#x27;t need that many changes.</p><p>Semantic Versioning defines a pattern similar to the one for alpha versions: <code>MAJOR.MINOR.PATCH-beta.NUMBER</code>, for example <code>2.0.0-beta.1</code>.</p><h4>Release candidate</h4><p>Release candidates (RC) are pre-release versions that you will want to publish for the general user-base to test an upcoming release. You want to publish a RC so that it&#x27;s battle tested by as many people as possible, in order to find possible bugs and edge cases.</p><p>They should not contain many bugs, but some could be found. The public API should not change anymore.</p><p>According to Semantic Versioning, the pattern to identify a RC is <code>MAJOR.MINOR.PATCH-rc.NUMBER</code>, for example <code>2.0.0-rc.1</code>.</p><h3>Conclusion</h3><p>We have seen some approaches and considerations to properly version your software.</p><p>Versioning is not so straightforward when you don&#x27;t have a lot of experience, but as you have seen, you just need to know a small set of rules and use a bit your intuition.</p><p>With those tools, you will be able to properly version your software.</p>

What to take into account when versioning software

<p>Two days ago I released the first major version in almost 2 years for <a target="_blank" rel="noopener noreferrer" href="https://github.com/shlinkio/shlink-web-client">shlink-web-client</a>, a web UI for <a target="_blank" rel="noopener noreferrer" href="https://shlink.io">shlink</a>, my self-hosted URL shortener.</p><p>This new version introduces several improvements to the project, and I thought it was worth writing a post to explain all of them in depth.</p><h3>New design</h3><p>The project has been using the same design almost since the beginning. It was clean, but it laked a bit of contrast and consistency in many sections.</p><p>Now the main background is light-gray, with components composed in white cards with a slight box shadow, that makes them easier to spot inside the app.</p><p>Also, the main area has now a max-width that prevents components from stretching too much on wide screens.</p><div class="row"><div class="col-md-6  "><img alt="Before" src="/assets/img/shlink-web-client-3/design-before.png" class="gallery-item"/><p class="text-center"><small>Before</small></p></div><div class="col-md-6  "><img alt="After" src="/assets/img/shlink-web-client-3/design-after.png" class="gallery-item"/><p class="text-center"><small>After</small></p></div></div><h3>Overview page</h3><p>Up until now, after connecting to a Shlink server, you went directly to the list of short URLs.</p><p>With this new version, you now see an overview page first, which displays some general stats, a quick short URL creation form, and the latest 5 short URLs.</p><p>From there, you can navigate to the full list of short URLs, with filtering capabilities, or go to the advanced short URL creation form.</p><div class="row"><div class="col-md-8 col-md-offset-2 "><img alt="Overview" src="/assets/img/shlink-web-client-3/overview.png" class="gallery-item"/><p class="text-center"><small>Overview</small></p></div></div><h3>Improved short URL creation form</h3><p>One of the sections with more room for improvement was the short URL creation form. It had all the inputs mixed, displaying just the &quot;long url&quot; one, with the option to un-collapsing the rest at will.</p><p>The new design groups the inputs by context, which makes it more clear. It also displays the inputs all the time, since the &quot;quick form&quot; is now located in the overview page.</p><div class="row"><div class="col-md-6  "><img alt="Before" src="/assets/img/shlink-web-client-3/creation-form-before.png" class="gallery-item"/><p class="text-center"><small>Before</small></p></div><div class="col-md-6  "><img alt="After" src="/assets/img/shlink-web-client-3/creation-form-after.png" class="gallery-item"/><p class="text-center"><small>After</small></p></div></div><h3>Enhanced visits page</h3><p>The visits page had a problem, it was trying to display too much information in a single place.</p><ul><li>Visits over time</li><li>Visits from every platform</li><li>Visits from every location</li><li>A list of visits as a table</li></ul><p>The new approach splits those in subsections with their own sub-route each (<code>/by-location</code>, <code>/by-context</code>, etc). This helps to render a smaller amount of charts at once, since they are a bit CPU greedy.</p><p>Also, only the visits from the last 30 days are loaded by default, making it load way faster for URLs with a lot of visits.</p><div class="row"><div class="col-md-6  "><img alt="Before" src="/assets/img/shlink-web-client-3/visits-before.png" class="gallery-item"/><p class="text-center"><small>Before</small></p></div><div class="col-md-6  "><img alt="After" src="/assets/img/shlink-web-client-3/visits-after.png" class="gallery-item"/><p class="text-center"><small>After</small></p></div></div><h3>Outer sections</h3><p>The app has some sections which are not related to a specific server, like creating/editing servers, or the settings page.</p><p>These sections have also been adapted to the new style, and they also have a max-width for wide screens.</p><div class="row"><div class="col-md-6  "><img alt="Before" src="/assets/img/shlink-web-client-3/create-server-before.png" class="gallery-item"/><p class="text-center"><small>Before</small></p></div><div class="col-md-6  "><img alt="After" src="/assets/img/shlink-web-client-3/create-server-after.png" class="gallery-item"/><p class="text-center"><small>After</small></p></div></div><h3>Date range selector</h3><p>A new component allows selecting date ranges either by relative time ranges (last 30 days, last 90 days...) or by selecting absolute start and end dates.</p><p>This component is used to filter both short URL lists and visits lists.</p><div class="row"><div class="col-md-6  col-md-offset-3"><img alt="Date range selector" src="/assets/img/shlink-web-client-3/date-range-selector.png" class=""/><p class="text-center"><small>Date range selector</small></p></div></div><h3>Domain selector</h3><p>Shlink has had multi-domain support for some versions, but changing the domain from the web client required to set the value manually, which is a bit error-prone.</p><p>This version introduces a new component which allows selecting from domains already used in the past (which is what you will want in most of the cases), preventing accidental typos.</p><p>It also allows setting new values in case you want to use a new domain that you never used before. This domain will then appear in the list the next time.</p><div class="row"><div class="col-md-6  "><img alt="Existing domains" src="/assets/img/shlink-web-client-3/domains-existing.png" class=""/><p class="text-center"><small>Existing domains</small></p></div><div class="col-md-6  "><img alt="New domain" src="/assets/img/shlink-web-client-3/domains-new.png" class=""/><p class="text-center"><small>New domain</small></p></div></div><h3>Other improvements</h3><p>The changes described so far were the major ones. However, v3.0.0 includes a few other minor improvements.</p><ul><li>Result and loading messages have been standardized: They are now more consistent and use the same set of components, which also ensures consistency in the future.</li><li>Error messages are now more descriptive: Instead of showing generic error messages when API calls fail, the actual error message is now always displayed, so that it&#x27;s easier to know what failed.</li><li>Dropped support for Shlink v1: All major versions come with some breaking changes that improve code maintainability. This one drops Shlink v1 support, which allows removing some conditional paths here and there.</li></ul><p>Other than the last point, Shlink v3.0.0 is mostly backwards compatible, so it&#x27;s safe to upgrade if you are already using Shlink v2.x</p><h3>Next steps</h3><p>Of course this doesn&#x27;t end here, and I have a lot of ideas to improve it even further.</p><ul><li>Dark theme: It&#x27;s what the cool kids do these days, and I have already done some experiments around it.</li><li>Shlink dashboard: Next gen web app for Shlink. You can read the whole reasoning in <a target="_blank" rel="noopener noreferrer" href="https://github.com/shlinkio/shlink-web-client/issues/338">this issue</a>.</li><li>Filtering in lists added to URL: Many sections allow filtering lists, but the state is saved in memory, which means it gets lost if the page is refreshed, and it cannot be bookmarked. The intention is to make all filtering params to become part of the URL.</li><li>Exporting visits in CSV: It&#x27;s cool to see charts full of info, but sometimes admins need to feed other services with the data, and consuming the original API is not always an option. Exporting visits to a CSV file will be supported soon by shlink-web-client.</li></ul>

Released shlink-web-client v3.0.0

<p>Let&#x27;s not deny that <a target="_blank" rel="noopener noreferrer" href="https://github.com/features/actions">GitHub actions</a> are hitting hard in the tech community, and many projects are transitioning from other continuous integration systems like <a target="_blank" rel="noopener noreferrer" href="https://circleci.com/">Circle CI</a> or <a target="_blank" rel="noopener noreferrer" href="https://travis-ci.org/">Travis CI</a>.</p><p>Personally I have been using Travis a lot during the past years, for all my OSS projects, and I have gotten very used to it.</p><p>However, people seem to be &quot;abandoning&quot; travis in favor of GitHub Actions, for different reasons. &quot;It&#x27;s faster&quot;, &quot;I can run smaller tasks in parallel&quot;, etc.</p><p>In my case, I have moved a few very specific tasks there, where travis was becoming a bottleneck, but I still have mixed feelings about the very-verbose yaml-based config needed to configure a few GitHub Actions, compared to the more concise travis one.</p><p>In this article I&#x27;ll explain how I improved a travis build to run several smaller tasks in parallel jobs, without having to completely move away from it.</p><h3>Initial state</h3><p>This project&#x27;s travis config is relatively simple. It is used to build a JS project for the web browser, and it has to run the next tasks:</p><ul><li>Lint the code, to make sure it fulfils the project&#x27;s coding standards.</li><li>Run unit tests.</li><li>Run mutation tests.</li><li>Build the docker image, to make sure none of the new changes broke it.</li><li>Conditionally publish a release on GitHub, with a dist file attached to it, if the build is being run for a git tag.</li></ul><pre><pre><code class="language-yaml">dist: bionic

language: node_js

branches:
  only:
    - /.*/

cache:
  directories:
    - node_modules

services:
  - docker

node_js:
  - &#x27;12.16.3&#x27;

install: npm ci

before_script:
  - echo &quot;Building commit range ${TRAVIS_COMMIT_RANGE}&quot;
  - export MUTATION_FILES=$(git diff ${TRAVIS_COMMIT_RANGE:-origin/main} --name-only | grep -E &#x27;src\/(.*).(ts|tsx)$&#x27; | paste -sd &quot;,&quot;)

script:
  - npm run lint
  - npm run test:ci
  - npm run mutate:ci
  - docker build -t project:test .

after_success:
  - node_modules/.bin/ocular coverage/clover.xml

before_deploy: npm run build ${TRAVIS_TAG#?}

deploy:
  - provider: releases
    api_key:
      secure: &lt;key&gt;
    file: &quot;./dist/project_${TRAVIS_TAG#?}_dist.zip&quot;
    skip_cleanup: true
    on:
      tags: true
</code></pre></pre><h3>Problems with this approach</h3><p>This is more or less ok, but the last two commands in <code>script</code> can be relatively slow and can delay the whole process.</p><p>Also, specially the mutation tests one, which is currently always considered successful regardless the result, can make the build reach the maximum run time allowed by travis, making the whole build falsely fail.</p><blockquote><p>The reason for the <code>before_script</code> step is to determine which source code files changed, and run mutation tests on them only, in order to mitigate this.</p></blockquote><p>Other than this, the build config defines &quot;global&quot; customizations that are actually only used for some tasks:</p><ul><li>Running <code>npm ci</code> is not required for the <code>docker build</code> task.</li><li>Getting the <code>docker</code> service is not required by any of the other tasks.</li><li>Determining changed files is only used by the <code>mutation tests</code> task.</li></ul><h3>Making use of jobs</h3><p>One cool feature travis supports is the use of a job <a target="_blank" rel="noopener noreferrer" href="https://docs.travis-ci.com/user/build-matrix/">build matrix</a>, that lets you define multiple jobs that run in parallel as part of the same build, and make each one of them do unrelated tasks.</p><p>Using this feature it is possible to improve the <code>.travis.yml</code> file, making it look like this:</p><pre><pre><code class="language-yaml">dist: bionic

language: node_js

branches:
  only:
    - /.*/

cache:
  directories:
    - node_modules

node_js:
  - &#x27;12.16.3&#x27;

jobs:
  fast_finish: true
  allow_failures:
    - name: &#x27;Mutation tests&#x27;
  include:

    - name: &#x27;Lint&#x27;
      install: npm ci
      script: npm run lint

    - name: &#x27;Unit tests&#x27;
      install: npm ci
      script: npm run test:ci
      after_success:
        - node_modules/.bin/ocular coverage/clover.xml

    - name: &#x27;Mutation tests&#x27;
      install: npm ci
      before_script:
        - echo &quot;Building commit range ${TRAVIS_COMMIT_RANGE}&quot;
        - export MUTATION_FILES=$(git diff ${TRAVIS_COMMIT_RANGE:-origin/main} --name-only | grep -E &#x27;src\/(.*).(ts|tsx)$&#x27; | paste -sd &quot;,&quot;)
      script: npm run mutate:ci

    - name: &#x27;Build docker image&#x27;
      services:
        - docker
      install: skip
      script: docker build -t project:test .

    - name: &#x27;Publish release&#x27;
      if: tag IS present
      install: skip
      script: skip
      before_deploy: npm run build ${TRAVIS_TAG#?}
      deploy:
        - provider: releases
          api_key:
            secure: &lt;key&gt;
          file: &quot;./dist/project_${TRAVIS_TAG#?}_dist.zip&quot;
          skip_cleanup: true
          on:
            tags: true
</code></pre></pre><p>This approach introduces all these benefits:</p><ul><li>Now linting, testing, mutation testing and building the docker image run in parallel, making the build finish earlier.</li><li>The mutation testing step is allowed to fail, and thanks to the <code>fast_finish</code> flag, travis won&#x27;t even wait for it to finish, considering the build successful once the rest have passed.</li><li>Configs that are only meaningful for certain jobs are defined in the job itself, instead of globally.</li><li>The &quot;publish release&quot; job only runs conditionally if the build is for a tag, thanks to the <code>if: tag IS present</code> condition. (More info about <a target="_blank" rel="noopener noreferrer" href="https://docs.travis-ci.com/user/conditional-builds-stages-jobs/">conditional jobs</a> in travis).</li><li>As soon as any of the jobs fails, the whole build fails and you don&#x27;t have to continue waiting. Before this, if the third task failed, you still had to wait for the previous two to finish.</li><li>Each job can have a name, making it more clear in travis UI what tasks are being run:</li></ul><p><img src="/assets/img/travis-multiple-jobs/travis-jobs.png" alt="Travis CI jobs"/></p><h3>Conclusion</h3><p>The main purpose of this article was to showcase that travis still has a lot to give. Github Actions are definitely cool, but there&#x27;s usually no silver bullet for everything, and a combination with other tools might be more suitable.</p><p>Also, this allows people using travis to improve their builds without having to change everything. Then, once you have split your tasks, if you still want to move to Github Actions, it should be easier.</p><blockquote><p>If you are curious, this is the project in which I use this travis configuration: <a target="_blank" rel="noopener noreferrer" href="https://github.com/shlinkio/shlink-web-client">shlinkio/shlink-web-client</a>.</p></blockquote>

Running several steps of a Travis CI build in parallel

<p>Asynchronous and non-blocking runtimes are pretty usual in many programming languages, as well as long-lived web apps that stay in memory and are capable of dispatching multiple HTTP requests without having to be fully bootstrapped every time. This has not been traditionally the case with PHP apps. </p><p>However, many projects are starting to get some adoption, that bring this long-lived approach to the PHP world.</p><p>The problem with this is that PHP developers are not used to this, and they tend to continue acting as they have always done. Also, libraries might have made some assumptions, making them not work as expected when used under this context.</p><p>Some of these projects are <a target="_blank" rel="noopener noreferrer" href="https://www.swoole.co.uk">swoole</a>, <a target="_blank" rel="noopener noreferrer" href="https://reactphp.org/">ReactPHP</a> or <a target="_blank" rel="noopener noreferrer" href="https://amphp.org/">Amphp</a>, to give some examples.</p><p>In this article, I will be focusing on the first one, and explain how to approach some of the pain points I have found after using it for a year and a half on an <a target="_blank" rel="noopener noreferrer" href="https://github.com/shlinkio/shlink">existing PHP project</a>.</p><h3>Introducing Swoole</h3><p>The main difference swoole has compared to other similar projects is that it&#x27;s written as a native PHP extension, and needs to be installed with <code>pecl</code>.</p><p>This is obviously a bit cumbersome, but the benefit is that, since it&#x27;s written in C++, it has a better handling of memory allocation, which is very important to achieve what it tries to do.</p><p>What swoole does is running a main process which bootstraps the PHP application once, and then keeps it in memory so that it can continue dispatching requests without having to load resources every time. This makes it really fast.</p><p>Those HTTP requests are handled by a set of child processes called &quot;web workers&quot;. Swoole forks the contents in memory for the main process to each one of the workers, so each worker has a &quot;copy&quot; of the application in memory.</p><p>Each worker is independent, but each one of them can only handle one request at a time, so you have to increase the number of workers if you expect a lot of concurrent traffic.</p><p>Other than web workers, swoole also has another set of child processes called &quot;task workers&quot;, which can be used to delegate long tasks at runtime to prevent web workers from getting blocked.</p><p>Here you can see a diagram of swoole&#x27;s architecture.</p><p><img src="https://www.swoole.co.uk/images/how-swoole-works.png" alt="How Swoole works"/></p><blockquote><p>More information can be found in <a target="_blank" rel="noopener noreferrer" href="https://www.swoole.co.uk/how-it-works">Swoole&#x27;s website</a>.</p></blockquote><h3>Problems to solve</h3><p>Now that we know how swoole works, let&#x27;s see some of the challanges I have faced in th past, which are usually not that ovbious when you are used to think in terms of &quot;blocking&quot; and &quot;short-lived&quot;.</p><h4>Database connections are kept open</h4><p>One of the first things you find out is that, since the application stays in memory, any database connection that has been opened by a worker (either web or task worker), will be kept open and reused in subsequent executions or HTTP requests served by the app.</p><p>This, which is in theory good, has a side effect, which is not obvious during development: <strong>the opened connection will eventually expire</strong>.</p><p>Since most of the popular database abstraction layers in PHP are coming from the times in which PHP was just a bootstrap-on-every-request technology, they never cared too much about expired database connections, as they would be opened again on next request.</p><p>However, when using swoole and the like, you have to implement some mechanism which takes care of reconnecting or gracefully recovering when this happens.</p><blockquote><p>I wrote an article explaining how I solved this on a middleware-based app served with swoole and using doctrine: <a href="/2019/11/04/how-to-properly-handle-a-doctrine-entity-manager-on-an-expressive-application-served-with-swoole/">How to properly handle a doctrine entity manager on an expressive application served with swoole</a></p></blockquote><p>Obviously, swoole (and also others) provides some tools to work with databases (like this <a target="_blank" rel="noopener noreferrer" href="https://www.swoole.co.uk/docs/modules/swoole-async-mysql-client">MySQL client</a>), which I&#x27;m sure do not have this problem (although, I haven&#x27;t personally tried this), but they are more limited than other well adopted libraries.</p><h4>In-memory caches don&#x27;t behave as expected</h4><p>It is a frequent practice to use local in-memory caches, like APCu, to improve applications performance in production.</p><p>Usually APCu is faster than other options, like redis or memcached, so it&#x27;s the choice when you only have one instance of your app running (meaning, you don&#x27;t have a cluster behind a load balancer).</p><p>However, even if you are running your app on just one server with swoole, because of the web worker system explained earlier, you actually have several copies of your app in memory, and each one of them will have a separated and not shared cache when using APCu.</p><p>This can lead to inconsistencies which are hard to debug, so you need to have it in mind.</p><p>Also, since the information that is being loaded is kept in memory anyway, you will not notice such a big improvement by using APCu or other in-memory caches.</p><blockquote><p>Notice that this problem does not happen when using distributed/centralized caching solutions, such as redis or memcached, as in that case, the caching technology is shared by all workers.</p></blockquote><h4>No need to optimize file including/requiring</h4><p>Many projects have complex configuration systems which are spread into several files. While this make it more maintainable, it also has a performance impact when all those files need to be loaded on every request.</p><p>Because of that, there are libraries like <a target="_blank" rel="noopener noreferrer" href="https://github.com/laminas/laminas-config-aggregator">laminas-config-aggregator</a> that can dynamically load config from many sources and, if some conditions are met, they can merge the result in a single bigger file that can be used in later requests.</p><p>In the case of swoole, as all the files will be kept in memory once loaded for the first time, there&#x27;s no need for this kind of optimization.</p><h4>Class autoloading happens once</h4><p>This point is very related with the previous one.</p><p>Composer offers a lot of optimization options so that class autoloading is faster in production. While you usually will still want to use some of those so that the first time a class is being loaded it is still reasonably fast, any subsequent hits will just use the class which is already in memory.</p><p>One of the composer flags you will probably don&#x27;t want use with swoole is <code>--apcu-autoloader</code>, because of the reasons mentioned two points above this one.</p><h4>Services should be truly stateless</h4><p>This is something that you should always do anyway, but it&#x27;s specially important when a service instance is kept in memory between HTTP requests.</p><p>Sometimes it&#x27;s easy to end up having some small piece of internal state on a service, where we save something as a side effect of a method call, to end up using it when another one of the public methods is invoked.</p><p>While this could seem harmless when the whole app is bootstrapped on every request, it can have very dangerous side effects when the app is served with swoole, because the instance will persist until another request is dispatched by the same web worker.</p><p>Just imagine you saved some user data on a service, and it ends up being served to another user on the next request. That won&#x27;t look good.</p><p>Because of this, make sure your services are truly stateless, and if for any reason that&#x27;s not possible, at least remember to have some mechanism in place that flushes the data just before finishing the request.</p><h4>Hot reloading for dev envs is not that good</h4><p>One thing we usually take for granted is that we can modify a PHP file and just by making a new request, the new code will be executed. However, when the code is kept in memory, it&#x27;s not that straightforward.</p><p>When you are developing a project which is served with swoole, you will have to restart the server every time to get your changes applied.</p><p>Swoole has a mechanism to do a light server reload, which is faster. You can find the docs in <a target="_blank" rel="noopener noreferrer" href="https://www.swoole.co.uk/docs/modules/swoole-server-reload">it&#x27;s website</a>.</p><p>The problem with this is that you need to implement the mechanism to invoke that when a file is changed, or use some library that does it for you.</p><p>Also, I have noticed that even calling that and seeing that the server is reloaded, does not always have the expected effect, and you have to still end up doing a hard reload. It&#x27;s something I don&#x27;t fully understand and I still need to investigate a bit further.</p><h3>Conclusion</h3><p>As it always happens when new technologies appear, it takes some time to get used to them, but I strongly suggest you to give a try to swoole or any other non-blocking runtime, because it makes your apps super fast.</p><p>Just remember to have all of the above in mind ;-)</p><p>That said, here you can find some libraries that will make your life easier when trying to serve existing projects with swoole, without having to couple with it and being able to continue using your favourite framework:</p><ul><li>IDE helper: <a target="_blank" rel="noopener noreferrer" href="https://github.com/swoft-cloud/swoole-ide-helper">https://github.com/swoft-cloud/swoole-ide-helper</a></li><li>Mezzio (formerly Zend Expressive): <a target="_blank" rel="noopener noreferrer" href="https://github.com/mezzio/mezzio-swoole">https://github.com/mezzio/mezzio-swoole</a></li><li>Symfony: <a target="_blank" rel="noopener noreferrer" href="https://github.com/k911/swoole-bundle">https://github.com/k911/swoole-bundle</a></li><li>Laravel: <a target="_blank" rel="noopener noreferrer" href="https://github.com/swooletw/laravel-swoole">https://github.com/swooletw/laravel-swoole</a></li></ul><p>Also, here you can find some more articles talking about this topic:</p><ul><li><a target="_blank" rel="noopener noreferrer" href="https://www.zend.com/blog/why-you-should-use-asynchronous-php">https://www.zend.com/blog/why-you-should-use-asynchronous-php</a></li><li><a target="_blank" rel="noopener noreferrer" href="https://samsonasik.wordpress.com/2020/04/05/using-swoole-in-mezzio-application-with-sdebug/">https://samsonasik.wordpress.com/2020/04/05/using-swoole-in-mezzio-application-with-sdebug/</a></li></ul>

Considerations when working with async PHP runtimes like swoole

<p>Some of you probably know that I have a pet project in which I like to work from time to time. This project is a self-hosted URL shortener called <a target="_blank" rel="noopener noreferrer" href="https://shlink.io">Shlink</a>.</p><p>Shlink is built using expressive as the base framework for the HTTP-dispatching task. A while ago, an expressive module was released to officially support serving expressive apps with <a target="_blank" rel="noopener noreferrer" href="https://www.swoole.co.uk/">swoole</a>, and I decided to include it on Shlink.</p><h3>How does swoole work?</h3><p>Swoole is an asynchronous non-blocking I/O framework which works in a similar way as node.js, but for PHP apps.</p><p>It has been conceived in a way that the app stays in memory between requests, removing the bootstrapping step and making apps to be served much faster.</p><p>Swoole can serve as many simultaneous requests as you tell it to. For this it makes use of workers. Each worker can serve one request at a time, and every worker has a separate instance of the app loaded in memory (this is something I <a target="_blank" rel="noopener noreferrer" href="https://twitter.com/dominikzogg/status/1189403552399134720">learned recently</a>).</p><p>Because of this, you have to sometimes change a bit the way you design your code. For example, if a service needs to be stateful (it shouldn&#x27;t, but shit happens), you will have to remember that the state will persist between requests, and you will want to somehow reset it to avoid unexpected side effects.</p><h3>How does this affect the EntityManager?</h3><p>One of the main services that keeps this internal state in Shlink is the doctrine <a target="_blank" rel="noopener noreferrer" href="https://www.doctrine-project.org/projects/doctrine-orm/en/current/tutorials/getting-started.html#obtaining-the-entitymanager">EntityManager</a>.</p><p>Due to the fact that it was originally designed to be used on a classical web server + CGI context, assuming it would be recreated on every request, there are a few considerations to take into account:</p><ul><li>It implements the unit of work pattern. It internally keeps track of all the entities that have been created/changed until you decide to flush it.</li><li>If something fails, the <code>EntityManager</code> can get closed, which makes it unusable after that moment.</li><li>If the database connection expires, the <code>EntityManager</code> will not gracefully recover. Instead, it will fail, since it&#x27;s not aware of the connection not being usable, getting yourself in previous point.</li></ul><h3>How to workaround it</h3><p>I learned about these issues by try and error, by using it on the app and getting the failures. Now, I want to share how I workaround those issues.</p><p>The first one I faced was the one making the app unusable after the <code>EntityManager</code> was closed. In order to solve this I created a decorated instance (naming, an object implementing <code>EntityManagerInterface</code> which in turn was wrapping an actual <code>EntityManager</code>), which was making sure the wrapped instance was transparently recreated if at some point it was closed.</p><p>However, this caused some side effects, and I finally realized that it was better to make the <code>EntityManager</code> to have to be reopened from the outside, and only once per request (this is closer to how it would behave on a classical set-up).</p><p>That&#x27;s how I came up with the <code>ReopeningEntityManager</code> Shlink is currently using (as of v1.20.0):</p><pre><pre><code class="language-php">&lt;?php

declare(strict_types=1);

namespace Shlinkio\Shlink\Common\Doctrine;

use Doctrine\ORM\Decorator\EntityManagerDecorator;

class ReopeningEntityManager extends EntityManagerDecorator
{
    /** @var callable */
    private $createEm;

    public function __construct(callable $createEm)
    {
        parent::__construct($createEm());
        $this-&gt;createEm = $createEm;
    }

    public function open(): void
    {
        if (! $this-&gt;wrapped-&gt;isOpen()) {
            $this-&gt;wrapped = ($this-&gt;createEm)();
        }
    }
}
</code></pre></pre><blockquote><p>You can find the actual code in <a target="_blank" rel="noopener noreferrer" href="https://github.com/shlinkio/shlink-common/blob/master/src/Doctrine/ReopeningEntityManager.php">shlink-common</a>.</p></blockquote><p>It just expects a factory that creates the actual <code>EntityManager</code>, and exposes an <code>open</code> public method which re-creates the wrapped instance if closed.</p><p>This method is called from a <a target="_blank" rel="noopener noreferrer" href="https://www.php-fig.org/psr/psr-15/">PSR-15</a> middleware which needs to be registered early in the middleware pipeline. It makes sure the method is called before the next middleware on the stack is invoked:</p><pre><pre><code class="language-php">&lt;?php

declare(strict_types=1);

namespace Shlinkio\Shlink\Common\Middleware;

use Psr\Http\Message\ResponseInterface;
use Psr\Http\Message\ServerRequestInterface;
use Psr\Http\Server\MiddlewareInterface;
use Psr\Http\Server\RequestHandlerInterface;
use Shlinkio\Shlink\Common\Doctrine\ReopeningEntityManager;

class CloseDbConnectionMiddleware implements MiddlewareInterface
{
    /** @var ReopeningEntityManager */
    private $em;

    public function __construct(ReopeningEntityManager $em)
    {
        $this-&gt;em = $em;
    }

    public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface
    {
        $this-&gt;em-&gt;open();

        try {
            return $handler-&gt;handle($request);
        } finally {
            $this-&gt;em-&gt;getConnection()-&gt;close();
            $this-&gt;em-&gt;clear();
        }
    }
}
</code></pre></pre><blockquote><p>Again, the code can be found in <a target="_blank" rel="noopener noreferrer" href="https://github.com/shlinkio/shlink-common/blob/master/src/Middleware/CloseDbConnectionMiddleware.php">shlink-common</a></p></blockquote><p>This middleware also does two more things (this time, after calling the next middleware on the stack, and making sure it does it even if an exception is thrown).</p><ul><li>It first closes the database connection. By doing it so, we make sure the <code>EntityManager</code> will reconnect the next time it needs to interact with the database, and therefore, we avoid the &quot;expired connection&quot; issue.</li><li>Then it clears the <code>EntityManager</code> itself. This avoids memory leaks, and also prevents that any orphan/non-flushed entity &quot;transcends&quot; to the next request, solving the first problem.</li></ul><h3>Further considerations</h3><p>This approach has been working well so far, but it&#x27;s important to assume that the <code>EntityManager</code> was not really designed for this non-blocking kind of context.</p><p>You need to know that the worker making use of it will not be able to serve any request until the <code>EntityManager</code> finishes its job. Because of this you might want to increase the number of swoole workers.</p><p>Also, closing the DB connection on every request is not the most optimal approach. It would be perfect to have a connection pool that was able to recreate connections when expired, and to keep them open as long as possible for faster operations. However, the <code>EntityManager</code> was not designed for that, so it&#x27;s better to close it.</p><p>You also need a <code>callable</code> to pass to the <code>ReopeningEntityManager</code>. The way you solve this depends on your implementation and the dependencies you use.</p><p>In my case, in Shlink I make use of the <code>ServiceManager</code>, so I have a delegator factory which takes the actual factory and passes it to the <code>ReopeningEntityManager</code>:</p><pre><pre><code class="language-php">&lt;?php

declare(strict_types=1);

namespace Shlinkio\Shlink\Common\Doctrine;

use Psr\Container\ContainerInterface;

class ReopeningEntityManagerDelegator
{
    public function __invoke(ContainerInterface $container, string $name, callable $createEm): ReopeningEntityManager
    {
        return new ReopeningEntityManager($createEm);
    }
}
</code></pre></pre><blockquote><p>You can find the actual code in <a target="_blank" rel="noopener noreferrer" href="https://github.com/shlinkio/shlink-common/blob/master/src/Doctrine/ReopeningEntityManagerDelegator.php">shlink-common</a>.</p></blockquote><h3>Conclusion</h3><p>I wanted to write this article as a recap for myself too, since I had to ask and investigate many times, and I couldn&#x27;t find a single source of truth explaining everything.</p><p>I hope that, if you ended up here, this article served you well.</p>