<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[TUI MM Engineering Center - Medium]]></title>
        <description><![CDATA[News, knowledge and technology at TUI Musement - Medium]]></description>
        <link>https://medium.com/tuimm?source=rss----a93729d9156b---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>TUI MM Engineering Center - Medium</title>
            <link>https://medium.com/tuimm?source=rss----a93729d9156b---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 10:27:23 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/tuimm" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How my team and I avoided flaky tests in TestCafe]]></title>
            <link>https://medium.com/@claudia.colacicchi1/how-my-team-and-i-avoided-flaky-tests-in-testcafe-664a55496fb5?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/664a55496fb5</guid>
            <category><![CDATA[testing]]></category>
            <category><![CDATA[quality-assurance]]></category>
            <category><![CDATA[e2e-testing]]></category>
            <category><![CDATA[testcafe]]></category>
            <category><![CDATA[flaky-tests]]></category>
            <dc:creator><![CDATA[Claudia Colacicchi]]></dc:creator>
            <pubDate>Fri, 04 Aug 2023 16:00:17 GMT</pubDate>
            <atom:updated>2023-08-04T16:00:17.408Z</atom:updated>
            <content:encoded><![CDATA[<h4>Have you ever struggled with flaky tests? Did it ever feel like “mission impossible”? Well, brace yourself: this article will take you through how my team and I drastically reduced test flakiness in <a href="https://testcafe.io/">TestCafe</a>!</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*PF4RZ6-0u0YobJTn" /></figure><blockquote>My name is <a href="https://www.linkedin.com/in/claudia-colacicchi-b02a1b15a/">Claudia Colacicchi</a>. I work as a Software Quality Assurance (QA) Specialist at <a href="https://www.tuigroup.com/en-en/about-us/about-tui-group/our-business/tui-musement">TUI Musement</a> and in my journey with testing automation I learned that not only are there amazing advantages of automating test scenarios but also that the main struggle is represented by <strong>flaky tests</strong>. For this reason, I thought it would be nice to share with you all how my team and I put in place a strategy against them!</blockquote><p>Although I am sure the majority of you know, I would like to start by defining what a flaky test is and why it is bad in Quality Assurance. <em>A flaky test is a software test that yields both passing and failing results despite no changes to the code or test</em>. The reason why having flaky tests is bad for a Quality Assurance Team is that this leads to instability and lack of confidence, which is a combo to absolutely avoid. Our job is to provide the product teams with a trusted level of confidence about their features and our test runs should always have the highest possible level of stability and reliability.</p><p>Let’s get to the juicy part though: how did we fix this problem? Our strategy consisted of three main approaches which I am going to describe one by one: custom retries inside TestCafe methods, test retries with TestCafe’s <a href="https://testcafe.io/documentation/403841/guides/intermediate-guides/quarantine-mode">Quarantine Mode</a>, and retrying missing tests.</p><h3>1- Custom retries inside TestCafe methods</h3><p>The very first flakiness we struggled with was our UI tests not being able to find web elements to interact with in the DOM. Because of this, our tests often did not go through some specific steps. We initially thought such elements were actually not displayed in the web page at times, however the screenshots showed them completely visible and also during our debugging sessions we saw them and could interact with them. We were clearly dealing with flakiness. We decided to address it with <strong>custom retries</strong> inside the specific functions that interacted with such web elements.</p><p>Here is an example. In the payment page of our <a href="https://www.tuiexperiences.com/uk/">e-commerce website</a>, there are fields to fill out with personal details before completing the purchase. One of them is the first name field where customers need to write their name. Well, for some reason, our tests often failed to validate the existence of this field. You can see an example of such failure below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VTk3ETej4UrQQ5TY246AZw.png" /></figure><p>The first approach we put in place to reduce the occurrence of this flakiness was a retry that <strong>executes the assertion three times before failing</strong>. To do so, as part of the <em>checkFirstNameField()</em> method, we implemented a <a href="https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Loops_and_iteration#for_statement">for loop</a> to allow for repeating the assertion. Inside the loop, we inserted an <a href="https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/if...else">if…else statement</a> to check whether the maximum number of desired attempts was reached and the repetition should stop. We set a retry limit to 4 and decided to throw an error in case the limit is reached, this means that if the assertion is repeated 3 times and still fails, the test fails and prints an error message saying <em>“Maximum number of attempts reached. The First Name field could not be reached”</em>. While the maximum number of attempts is not reached, the function keeps repeating the assertion in case it fails, because it enters the else block and executes a <a href="https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/try...catch">try…catch statement</a>. The assertion to verify if the first name field exists is performed inside the try block, if it succeeds then the <em>firstNameFieldIsAvailable</em> variable is set to true, the for loop is exited and the test can continue on to the next method. If the assertion fails, the catch block is entered, the <em>firstNameFieldIsAvailable</em> is set to false and the loop continues. Below, you can see the related snippet of code:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CYk8677oLz2uAcXVwCrqkw.png" /></figure><p>We had to apply these custom retries on top of an already present built-in feature by TestCafe which performs retries itself: the <a href="https://testcafe.io/documentation/402827/guides/advanced-guides/built-in-wait-mechanisms"><strong>Built-in Wait Mechanism</strong></a>. Let me explain better what it is and why, despite that, we had to come up with a customized solution to fix our problem. In the case of an assertion, which is our example, TestCafe executes the <em>Smart Assertion Query Mechanism </em>which does not statically wait for page elements to appear but repeats the evaluation process until the assertion yields a successful result, or the <a href="https://testcafe.io/documentation/402639/reference/command-line-interface#--assertion-timeout-ms">assertion timeout</a> expires. Was this not enough for preventing our flakiness? Unfortunately not.</p><p>You might be wondering why we did not just extend the assertion timeout and let TestCafe do its intelligent job behind the scenes with a little extra time. Well, we did, but it was not enough to have significantly stable tests: until we implemented our custom retries, we still faced a certain degree of flakiness which we could not afford. The really nice thing is that, right now, we have a customized retying mechanism implemented on top of a smart built-in feature by TestCafe that repeats evaluating if the element exists inside each of our custom retry. Pretty nice combination, huh?</p><h3>2- Test retries with TestCafe’s Quarantine Mode</h3><p>In certain cases, implementing retries within TestCafe’s methods was not enough. At this point, we decided to strengthen our strategy by making use of a very nice functionality powered by TestCafe: <a href="https://testcafe.io/documentation/403841/guides/intermediate-guides/quarantine-mode"><strong>Quarantine Mode</strong></a>.<br>Quarantine Mode is a feature aimed at reducing false negatives and instability. When a test fails, TestCafe will quarantine it and will <strong>repeat it until considered successful</strong>.</p><p>How can TestCafe eventually confirm that the test is successful? For this scope, TestCafe provides a configurable option named “<strong>success threshold</strong>”, which is the number of successful attempts necessary to confirm the test’s success. The value of this “success threshold” can be decided by the QA Team. My team and I agreed to set it to 1 and it has proven to be reliable.</p><p>Are you now wondering what happens if the test is rightly failing because of a bug? Will TestCafe keep repeating the test forever because the success threshold is not met? Of course not! There is another option that covers this need and it is called “<strong>attempt limit</strong>”: it is the maximum accepted number of test attempts. Again, its value can be decided by the team, and in our case we set the attempt limit to 3, making TestCafe repeat our failed test 3 times at most and then mark it as failed in case it does not succeed once during the retries.</p><p>Below, you can find our own configuration of Quarantine Mode. To summarize, by setting it this way, TestCafe’s Quarantine Mode quarantines a failed test and repeats it for a maximum of 3 times (the attempt limit). Within these repetitions, if the test reaches a success threshold of 1, meaning that it succeeds at least once, TestCafe will mark it as passed, otherwise it will consider it failing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rP2dT7XJIa9f9tUHzNLGYA.png" /></figure><blockquote>Be aware that if the success threshold is met before reaching the attempt limit, the test will be marked as passed and TestCafe won’t keep executing it.</blockquote><h3>3- Retrying missing tests</h3><p>On top of false negatives and general problems making our tests wrongly fail, my team and I faced issues with tests not running at all. Sometimes, our reports showed incomplete test runs, where one or more test scenarios did not run.</p><blockquote>You could argue that having incomplete test runs is not a real test flakiness, which is true. However, we considered it fitting with the whole definition of having our tests fail to produce the same outcome with each individual run, which caused unreliability. That is why we included a solution to this problem in our strategy to tackle flakiness.</blockquote><p>We addressed this issue by implementing a check to verify which test is not present in the report, if any, and re-trigger its run. Below, you can see the script that helped us do that, and since our reporting system is <a href="https://qameta.io/allure-report/">Allure</a>, you will see references to it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wf7ZaFdb5-VmaLA17x_Xxg.png" /></figure><p>Since applying the three described solutions, my team and I saw a <strong>significant improvement</strong> in our test runs. We have many less false negatives than before and the reliability of our testing strategy increased greatly. We have never had an incomplete test run again and we have gained much more <strong>trust</strong> from our development teams. It was an amazing achievement that we reached as a team.</p><p>For this reason, I want to express my sincere gratitude to my team leader <a href="https://www.linkedin.com/in/aur%C3%A9lien-lair-844385104/">Aurélien Lair</a> and my teammates <a href="https://www.linkedin.com/in/andrea-zito-az/">Andrea Zito</a> and <a href="https://www.linkedin.com/in/david-contreras-sanchez-matamoros/">David Contreras</a> for their continuous support, their effort and passion to achieve our shared goals. This article would not exist if it was not for our amazing teamwork, thank you guys!</p><blockquote>We really hope that the solutions described in this article can be helpful for you Quality Assurance Engineers out there. In case you are struggling with the same issues we faced, we strongly suggest trying out our solutions!</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=664a55496fb5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[We built a dual track engineering career ladder. Here’s why.]]></title>
            <link>https://medium.com/tuimm/we-built-a-dual-track-engineering-career-ladder-heres-why-a767f60dc257?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/a767f60dc257</guid>
            <category><![CDATA[engineering-culture]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Umberto Di Fabrizio]]></dc:creator>
            <pubDate>Thu, 03 Mar 2022 15:30:15 GMT</pubDate>
            <atom:updated>2022-03-03T15:30:15.139Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n542Y5DcpmNTYTk5v8LqvQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@austindistel">Austin Distel</a></figcaption></figure><p>In my past experience as a software engineer I’ve had mixed feelings about the career ladder definition: more often than not the expectations for each role were hard to grasp. Now, as Head of Engineering at TUI Musement it’s my responsibility to craft a clear and transparent growth plan that can be used as a reference for our engineering department.</p><p>Our <strong>dual track career path</strong> allows engineers to evolve in two directions: Domain Expert or Management. The first one (in green) focuses on becoming an expert in specific technologies and broadening the scope of one’s impact, the latter (in red) aims at helping engineers to grow and lead teams to deliver high-quality solutions.</p><p>We believe that climbing the ladder doesn’t have the same end goal for everyone. In fact, being a great software engineer doesn’t mean you’ll be a great leader, so it accounts for two different types of career evolution.</p><blockquote>Our objective is to make sure that everyone has a path that allows tailored growth.</blockquote><p>A dual track protects engineers from making a leap towards the unknown, offering instead a clear path ahead and a picture of what’s behind each step in one direction or the other.</p><h3>⚠️ engineering != engineering_management</h3><p>Once you understand how and why management and engineering are two different jobs, the idea of two parallel tracks becomes quite obvious, but I learnt this truth the hard way 😔.</p><p>Early in my career, I was puzzled by the idea of having a manager doing ‘things’ other than coding with the team. I was surprised when <a href="https://www.amazon.com/Managers-Path-Leaders-Navigating-Growth/dp/1491973897">reading about career growth</a> that would explain how the two roles (crafting code and leadership) are different and require a different skillset. After all, as an engineer, I want my manager to be the best person to tell me how to move forward when stuck with a bug and to teach me the best practices. Yes, this is part of the role, but not even the hardest.</p><p>What does a manager do? Alignment, synchronization, conflict resolution, adds/removes processes. That’s it. There’s no Vue.js, C++, CQRS mentioned in the requirements.</p><blockquote>A manager is the result of company growth, it’s the glue that avoids chaos.</blockquote><p>Of course,<strong> you must be a great engineer to be an engineers’ leader, but that is not enough, </strong>you’ll have to update your skills toolbox 🛠</p><h3>The dual track</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oWj_q35y4hcYoAS0HQK_-g.png" /><figcaption>Dual track engineering ladder @ TUI Musement</figcaption></figure><p>The dual career path theme has been <a href="https://www.theengineeringmanager.com/management-101/the-two-tracks-of-growth">discussed a lot</a> in recent years and there are many companies that support this type of career growth and <a href="https://www.progression.fyi/">quite a number of inspirations</a>.</p><p>At TUI Musement software engineers grow (left to right) within the Domain Expert track until they reach the senior level. At that point in their career there are three possible choices:</p><ol><li>Follow the Domain Expert track, but amplify the scope of responsibility.</li><li>Change job and start the Management Track.</li><li>Stay at the senior level position (sometimes this choice is overlooked).</li></ol><p>While we expect every engineer to eventually reach the senior level, there is no expectation to grow above that unless there’s the will to do so.</p><p>In the sections below I will highlight the expectations for each of the two tracks.</p><h3>💻 Domain Expert Track</h3><ul><li>Responsible for <strong>implementing high-quality software</strong>.​</li><li>Deeply cares, advocates, and <strong>applies testing strategies</strong> to guarantee bug-free code.​</li><li>Constantly improves the codebase by crafting new algorithms, <strong>refactoring, </strong>and participating in code reviews.​</li><li>Designs scalable, maintainable, and <strong>observable solutions</strong>.</li></ul><p>If you want to know more about working as a Software Engineer in TUI MM you should check <a href="https://medium.com/u/2775d7a95ddb">Gionatan Lombardi</a> article 👇👇</p><p><a href="https://medium.com/tuimm/a-day-in-the-life-of-a-tui-musement-software-engineer-89c337923740">A day in the life of a TUI Musement Software Engineer</a></p><h3>👩‍👩‍👧‍👦 Management Track</h3><ul><li><strong>Regularly adjusts processes</strong> and timelines to ensure high-quality software is delivered.​</li><li>Improves team’s velocity and impact.​</li><li>Responsible for <strong>People Management </strong>(career conversations, performance management, objectives, 1:1, coaching, hiring…)​​</li></ul><p>When taking this path, one should be comfortable with the idea of:</p><ol><li>Not writing code</li><li>Leading people with a different skillset</li></ol><blockquote>Above all, the first responsibility of a manager is caring about the people he/she leads.</blockquote><p>Now you know the dual track system and the expectations for each track, but how can you get there?</p><h3>📈 Tools to climb the ladder</h3><p>Unfortunately, there’s no recipe to move up the ladder, except for: <strong>always be learning</strong>. The best advice I can give to engineers looking to move up the ladder is: act as if you’re already at the next level and don’t hold back because “that is not my role”. Look at what fellow engineers at the next level are doing and try to do the same.</p><blockquote>Act as if you’re already at the next level.</blockquote><p>There’s one last piece of the puzzle. We believe that<strong> the best way to move forward is by having constructive feedback, </strong>so never lose an opportunity to get mentored and to get advice.</p><blockquote>Growth requires high quality feedback.</blockquote><p>For this reason, we have created two tools in TUI Musement to guide career conversations:</p><ol><li>The TUI Musement roles’ expectations cards. <a href="https://tuigroup-my.sharepoint.com/:w:/g/personal/umberto_difabrizio_tui_com/ETGo72rQQRZNiJ1qp3zFmswBWd6LOhNdmPrzWkouXE003A">Here</a>, you can find a detailed explanation for each role. It’s important to consider that <strong>each card isn’t a recipe but an ideal state.</strong></li><li>The TUI Musement engineering competency matrix. This tool is used to evaluate 27 specific competencies (hard skills and soft skills) at each level of the track and to identify gaps. (Follow our blog, we’re planning a deep dive on this one 😉)</li></ol><p>✏️ <em>PS. you can use the roles cards to decide what role fits you best when applying for our </em><a href="https://careers.tuigroup.com/jobs/"><em>open positions</em></a><em> </em>😉.</p><p>Most people consider growth within an organization as getting promotion after promotion to get to the top. Yet, it can happen that once you find yourself in a particular role you understand that it does not fit with your aspirations, what can you do then?</p><h3>Lateral growth</h3><p>Choosing a career path between Domain Expert and Management isn’t a <a href="https://www.inc.com/jeff-haden/amazon-founder-jeff-bezos-this-is-how-successful-people-make-such-smart-decisions.html">one-way door</a>.</p><blockquote>Management is a different job, thus it’s possible you’ll miss the dopamine hit of fixing something or solving something.</blockquote><p>There’s value in going back and forth between Domain Expert and Management (<a href="https://charity.wtf/2017/05/11/the-engineer-manager-pendulum/">the pendulum</a> and <a href="https://charity.wtf/2019/01/04/engineering-management-the-pendulum-or-the-ladder/">sequel</a>) and the move it’s always a tradeoff (you’ll lose some responsibilities and gain others). We try to make these lateral movements as easy as possible, but it’s true that whilst moving from a Domain Expert role to a Management role it’s frequent, the opposite can be more challenging.</p><p>At this point, by knowing our framework for engineering growth and how we operate it, you might be wondering…</p><h3>Should you build a dual track at your company?</h3><p>Don’t take advice from the internet people. Yet.</p><ul><li>Are you a startup with less than 10 engineers? No. You’ve got much more important things to figure out and the hype is fuelling everyone’s brains.</li><li>More than 15? Start playing around with a few experiments.</li><li>More than 50? Let me know how you’re doing without.</li></ul><h3>Conclusion</h3><p>By now, I hope I’ve convinced you that a dual track ladder is the perfect tool to support engineers as they grow within the organization, and to set expectations about each level. At TUI Musement we adopt roles’ expectations cards to define clearly and precisely the required skills for each role, and we use a competency matrix to understand what are the skills gaps, and when one is ready to step up to a new role.</p><p>I hope that this article has opened a little window on our engineering culture, and that it is going to be inspiring for anyone looking to create a similar framework. If you’d like to have a chat about this or you’re interested in any of our <a href="https://careers.tuigroup.com/jobs/">open positions</a> feel free to reach me out on <a href="https://www.linkedin.com/in/umbertodifabrizio/">Linkedin</a>.</p><p>Kudos to <a href="https://medium.com/u/e938aba06441">Cristian Ronzio</a> for advocating this engineering culture and to all the reviewers that have made this article so much better with their suggestions ❤️.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a767f60dc257" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/we-built-a-dual-track-engineering-career-ladder-heres-why-a767f60dc257">We built a dual track engineering career ladder. Here’s why.</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Implement a UI library with Web Components]]></title>
            <link>https://medium.com/tuimm/implement-a-ui-library-with-web-components-b128e82f1f2a?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/b128e82f1f2a</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[storybook]]></category>
            <category><![CDATA[web-components]]></category>
            <category><![CDATA[stenciljs]]></category>
            <category><![CDATA[ui-library]]></category>
            <dc:creator><![CDATA[Marco Ziliani]]></dc:creator>
            <pubDate>Thu, 03 Mar 2022 15:30:05 GMT</pubDate>
            <atom:updated>2022-03-03T15:30:05.266Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-1VD0SANNsqpwH6r2xOafg.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/photos/gcHFXsdcmJE">Kelly Sikkema</a> on <a href="https://unsplash.com/">Unsplash</a></figcaption></figure><p>Recently in TUI Musement we introduced a <a href="https://stenciljs.com/docs/what-is-design-system">design system</a> shared by different projects. We had two main objectives: on one side we wanted to offer our users a more cohesive user experience across our platforms; on the other side we wanted to make our development flow faster and more scalable.</p><p>A design system is composed of many elements: a UI component library, a style guide (colours, fonts, etc), organisation principles, best practices, and more. While the designers started creating this complex system, the developers started focusing on how to implement the UI component library.</p><p>This is our starting point: we have three single page applications, created at different times by different people. One is built with <a href="https://reactjs.org/">React</a>, while the others are built with <a href="https://vuejs.org/">Vue.js</a> and <a href="https://nuxtjs.org/">Nuxt.js</a>. In addition, we know that the design system could be used for other apps in the future, so we don’t want to be tied to a specific framework.</p><p>In this article we are going to discuss how we approached the implementation of a component library and the problems we faced, focusing on these aspects:</p><ul><li>the technology we chose</li><li>how we are distributing the library</li><li>how to use the components in Vue and React</li><li>how to document the library with <a href="https://storybook.js.org/">Storybook</a></li></ul><h3>Web Components</h3><p>We thought about different solutions, but we soon convinced ourselves that the best choice is <a href="https://developer.mozilla.org/en-US/docs/Web/Web_Components">Web Components</a>: they are framework-agnostic, web standard and are now <a href="https://caniuse.com/?search=web%20components">supported by all modern browsers</a>.</p><p>While web components are convenient, it’s not easy to write one from scratch, and it requires a lot of boilerplate. So we decided to use a library to develop our components and we picked <a href="http://stenciljs.com/">Stencil.js</a>: it generates standards-compliant web components and <a href="https://stenciljs.com/docs/stencil-for-design-systems">it has been designed for design systems</a>. Moreover, it has a great build system, <a href="https://webcomponents.dev/blog/all-the-ways-to-make-a-web-component/">it’s reasonably lightweight</a>, and <a href="https://stenciljs.com/docs/my-first-component">it’s easy to use</a>.</p><h3>Example of a component</h3><p>As a <em>proof of concept</em>, we created a simple component, a button. We are not going to show the implementation, but you can find it <a href="https://github.com/zi/ui-library-poc/blob/master/src/components/button/my-button.tsx">here</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wdDPeyutHa7_A78eQ0SebQ.png" /></figure><h3>Distributing and importing the library</h3><p>One of the most interesting features of Stencil is the ability to generate different builds at the same time, letting the consumer decide how to import and use the components.</p><p>The first way to import the library is called <a href="https://stenciljs.com/docs/distribution"><em>distribution</em></a><em>. </em>It’s the easiest way for the consumer to integrate the library and it comes with two great features out of the box: lazy-loading and polyfills for legacy browsers (e.g. Internet Explorer 11). All the consumer has to do is include a tiny script and Stencil will load any component on demand. There’s no need to manually define <a href="https://developer.mozilla.org/en-US/docs/Web/Web_Components/Using_custom_elements">custom elements</a>.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8390d340f560c2a6ffa505ea233b6560/href">https://medium.com/media/8390d340f560c2a6ffa505ea233b6560/href</a></iframe><p>The second way is the <a href="https://stenciljs.com/docs/custom-elements"><em>Custom Elements Bundle</em></a><em>.</em> In this case Stencil will generate a single, <a href="https://developers.google.com/web/fundamentals/performance/optimizing-javascript/tree-shaking">tree-shakeable</a>, bundle, containing all the components. When using this bundle, the consumer must apply polyfills if needed and manually define every custom element.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/1c917a4234a0b3e144265d4f56a1af2e/href">https://medium.com/media/1c917a4234a0b3e144265d4f56a1af2e/href</a></iframe><p>On the other hand, by manually importing and defining the components, the consumer has more control on how the components are loaded.</p><p>Since our applications implement route-level code splitting and supports lazy-loading, we already have the ability to include only the code of the components that are actually used on the page. Therefore we decided to go with this method.</p><h4>The problem with code splitting</h4><p>Our idea was to import and define the custom elements in the page in which they were used.</p><p>We faced a problem though. If we are using the same component on page A and page B, and the user visits both pages, the custom element will be defined two times. Unfortunately, this is not allowed. The method CustomElementRegistry.define() <a href="https://developer.mozilla.org/en-US/docs/Web/API/CustomElementRegistry/define#exceptions">will raise an exception</a> if the custom element has already been defined.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/684/1*gbvclB-c_we1GWG6FRBqbA.png" /><figcaption>Error: Failed to execute ‘define’ on ‘CustomElementRegistry’: <br>the name “my-button” has already been used with this registry</figcaption></figure><p>To solve this problem we created a little utility to check if the custom element has already been defined:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/06ddd67112a42dece6f6b82784f718b7/href">https://medium.com/media/06ddd67112a42dece6f6b82784f718b7/href</a></iframe><p>To include this utility in our build, we created a <a href="https://github.com/zi/ui-library-poc/blob/master/rollup.config.js">rollup configuration</a> and added an extra step in the build phase:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/dc9f7fd713d5ef32c657bf08aa881a17/href">https://medium.com/media/dc9f7fd713d5ef32c657bf08aa881a17/href</a></iframe><p>Our consumer can import and define the custom element like this:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/3bd08b6e4902fd10166a79fc90514ffe/href">https://medium.com/media/3bd08b6e4902fd10166a79fc90514ffe/href</a></iframe><h3>Using web components in Vue</h3><p>Vue offers a good support to custom elements and integrating them in Vue is pretty straightforward.</p><p>To pass props to a custom element you can use <a href="https://vuejs.org/v2/api/#v-bind">binding</a> like with a normal Vue component. The only thing to keep in mind is the difference between <a href="https://javascript.info/dom-attributes-and-properties">HTML attributes and DOM properties</a>.</p><p>By default Vue uses attributes to bind props to a custom element. Since attribute values are always string, our props are converted to strings.</p><p>To pass complex data like arrays, objects or functions we can use the modifier .prop . Using this modifier the prop is bound as a DOM property instead of an attribute.</p><p>Vue can listen to native DOM events dispatched from our custom element in the way we are used to.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/44acf3ce5f65eedb449c749109f6d6ce/href">https://medium.com/media/44acf3ce5f65eedb449c749109f6d6ce/href</a></iframe><h3>Using Web Components in React</h3><p>Unfortunately <a href="https://reactjs.org/docs/web-components.html">React’s support to custom elements</a> isn’t as good as Vue’s. React passes all data to custom elements as HTML attributes, making it impossible to pass complex data (that is, data that is not a string or a number). Native custom events are not handled correctly.</p><p>A solution to these problems exists. We can obtain a ref to the custom element; we then use the ref to set properties on the element as DOM properties and to add event listeners via addEventListener.</p><p>We created a custom hook that can be used to handle this kind of logic for any custom element:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/192298dc4a7f567cb0afb7c994418e54/href">https://medium.com/media/192298dc4a7f567cb0afb7c994418e54/href</a></iframe><p>The hook takes three arguments:</p><ul><li>a ref to the custom element</li><li>an object containing the properties</li><li>an object containing event handlers</li></ul><p>Here is an example of how to use it:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/c679483a767ee2be02eca50a5a358a58/href">https://medium.com/media/c679483a767ee2be02eca50a5a358a58/href</a></iframe><h3>Setting up the Storybook</h3><p>We decided to use <a href="https://storybook.js.org/docs/web-components/get-started/introduction">Storybook</a> for two main reasons:</p><ul><li>it makes it easier to develop components: you can view the components in isolation and test them with different properties</li><li>it offers a great documentation, that can be shared between developers and designers</li></ul><p>In addition, it supports many frameworks (and web components, of course) and has an ecosystem of <em>addons</em> that extends its functionalities.</p><p>You can install Storybook in your existing project by running:</p><pre>npx sb init --type web_components</pre><p>To enable HMR (Hot Module Replacement) for web components you can follow <a href="https://github.com/storybookjs/storybook/tree/master/app/web-components#setup-page-reload-via-hmr">this documentation</a>.</p><h4>Development using Stencil and Storybook</h4><p>Both Stencil and Storybook provide a command to start the development server. We decided to create a command to run both using <a href="https://www.npmjs.com/package/concurrently">concurrently</a>:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/93e0d0e11bc566b6bbb13207f6a1c347/href">https://medium.com/media/93e0d0e11bc566b6bbb13207f6a1c347/href</a></iframe><p>When we run dev, Stencil builds the library into the folder dist. <a href="https://storybook.js.org/docs/web-components/api/cli-options#start-storybook">The option</a> -s in the storybook command tells Storybook to load static files from thedist folder.</p><p>We can now <a href="https://storybook.js.org/docs/web-components/configure/story-rendering#adding-to-head">add our library scripts into the head</a> of the Storybook page by creating the file .storybook/preview-head.html</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/554281e46be218d3615f015b540a9379/href">https://medium.com/media/554281e46be218d3615f015b540a9379/href</a></iframe><p>To view the component we are working on we can write a <a href="https://storybook.js.org/docs/web-components/writing-stories/introduction">story</a>:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/f00ffb4ec545a0fffe18284e4ddaa35d/href">https://medium.com/media/f00ffb4ec545a0fffe18284e4ddaa35d/href</a></iframe><p>This how the story looks in Storybook:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*d7k4nsDU5AAcUEcdzqt77w.png" /></figure><h4>Components documentation</h4><p>When you write stories during development you are also creating a <a href="https://storybook.js.org/docs/web-components/writing-docs/introduction">documentation</a> for your components.</p><p>The addon <a href="https://github.com/storybookjs/storybook/tree/next/addons/docs/web-components">Docs</a>, installed with Storybook, shows a documentation of properties and events defined in our component.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HgZtdxzrHNbRBJTdrLm4bQ.png" /></figure><p>As explained in <a href="https://github.com/storybookjs/storybook/tree/9d4525b4592af6fa49fc7d68bd004849159048f3/addons/docs/web-components#props-tables">the addon’s documentation</a>, the plugin needs a json file that describes the custom elements.</p><pre>import { setCustomElements } from &#39;@storybook/web-components&#39;;<br>import customElements from &#39;../custom-elements.json&#39;;</pre><pre>setCustomElements(customElements);</pre><p>This file, custom-elements.json, can be generated automatically.</p><p>Unfortunately, the method to generate it explained in the documentation is not updated and is no longer working.</p><p>There is an alternative, as indicated <a href="https://github.com/ionic-team/stencil/issues/2328#issuecomment-810474959">here</a>. Stencil exposes an <a href="https://stenciljs.com/docs/docs-custom">output target</a> that generates the docs in the form of json data that we can modify.</p><p>We can adapt the automatically generated docs to create our custom-elements.json like this:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/ecb2cd446e57dbbe158044a1699461df/href">https://medium.com/media/ecb2cd446e57dbbe158044a1699461df/href</a></iframe><p>Adding the docs-customoutput target in stencil.config.ts, when we build the library with the flag --docs, Stencil will also update custom-elements.json.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/077eb7a6ed71ac2c98b2a40904d76d0e/href">https://medium.com/media/077eb7a6ed71ac2c98b2a40904d76d0e/href</a></iframe><h3>Conclusion</h3><p>We are very satisfied with our UI component library so far, and with the tools we are using to build it.</p><p>Stencil provided a good developer experience, making it easy to write standard web components, and provided great build tools to distribute them; Storybook is the perfect companion, both for development and for documentation.</p><p>It can also reduce the distance between designers and developers, thanks to its <a href="https://medium.com/storybookjs/storybook-for-design-3ff761c579bf">integrations</a>. The improved collaboration is one of the best outcomes of implementing a shared component library.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b128e82f1f2a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/implement-a-ui-library-with-web-components-b128e82f1f2a">Implement a UI library with Web Components</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Elevate Your Web App With Augmented Reality]]></title>
            <link>https://medium.com/tuimm/elevate-your-web-app-with-augmented-reality-350bef8a44f2?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/350bef8a44f2</guid>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[augmented-reality]]></category>
            <category><![CDATA[engineering-lab]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Gabriele Fazio]]></dc:creator>
            <pubDate>Thu, 03 Mar 2022 15:29:38 GMT</pubDate>
            <atom:updated>2022-03-03T15:29:38.837Z</atom:updated>
            <content:encoded><![CDATA[<h4><a href="https://en.wikipedia.org/wiki/Augmented_reality">AR</a> could be the next big thing, here’s an example of how you can use it to improve user experience</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CjYDwOD9Rp8xehPXn7BkaQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@patrick_schneider?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Patrick Schneider</a> on <a href="https://unsplash.com/s/photos/augmented-reality?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>2020 was an exceptional year. The pandemic has profoundly changed our habits and the way we approach technology. From work, education to leisure and shopping, every area of our lives has been affected. In this new digital world, there are plenty of opportunities for us as developers to engage with our users and to provide innovative and exciting ways to experience the web. One of these is Augmented Reality, a technology that has seen unprecedented growth in the last few years, with a market that is worth approximately $15.3 billion, and is forecast to reach $72.8 billion by 2024.</p><h3>What is AR?</h3><p>Well, if you have ever tried an <a href="https://www.instagram.com/">Instagram</a> or <a href="https://www.snapchat.com/l/en-gb/">Snapchat</a> filter, or if you are a bit of a nerd like me, and you’ve been running around your town trying to catch your favourite Pokemon, there’s a high chance you already know what AR is about. To put it in a few words:</p><blockquote>Augmented Reality is a layer of digital objects and information, rendered over the real world</blockquote><p>What’s the difference between <a href="https://en.wikipedia.org/wiki/Augmented_reality">Augmented Reality</a> and <a href="https://en.wikipedia.org/wiki/Virtual_reality">Virtual Reality</a>, I hear you ask? AR allows the user to be in control of the environment, which has been digitally augmented or enhanced in some way. VR, on the other hand, is a fully immersive experience, where the real-world is replaced with a digitally simulated one.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*djGbEi5zt4u96lSbN7vFSQ.png" /></figure><h3>…and why should we care?</h3><p>First of all, AR can simplify user’s life. An example would be IKEA Place, an app that lets you virtually place furniture directly in your room.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2FrdIBv87b0zMFa%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fgiphy.com%2Fgifs%2Fadweek-furniture-ar-rdIBv87b0zMFa&amp;image=https%3A%2F%2Fmedia2.giphy.com%2Fmedia%2FrdIBv87b0zMFa%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="244" frameborder="0" scrolling="no"><a href="https://medium.com/media/dcd728b40ac563c431ff9951e61cc64e/href">https://medium.com/media/dcd728b40ac563c431ff9951e61cc64e/href</a></iframe><p>Multiple retailers are now implementing apps that let users try their product online using AR technology (try <a href="https://www.glasses.com/gl-us/virtual-try-on">this</a>, for example). You can see how powerful AR can be in this case. The user no longer has to guess how a specific product would look before buying it, and this builds trust and confidence, which translates to higher conversion rates.</p><p>Another important aspect of AR is that it immediately captures the user’s attention, making it more likely that he/she will spend more time interacting and engaging with your product.</p><h3>AR for the web</h3><p>As of February 2021, AR and VR are accessible from most modern browsers, thanks to technologies like <a href="https://webassembly.org/">Web Assembly</a> and <a href="https://developer.mozilla.org/en-US/docs/Web/API/WebGL_API">WebGL</a>. These technologies are used to leverage some of the device’s features (Camera, GPS, Sensors) to render 3D objects and to create interactive experiences. At the moment, WebAR only offers a limited selection of the main features possible using native app AR, but it has the overwhelming benefit of being accessible by most devices, without having to buy expensive equipment. With an app, the size or data allowance for download and device type can discourage people to use AR, but WebAR makes it more accessible and doesn’t eat up people’s data through chunky downloads.</p><blockquote>Note: At the time of this article, the new <a href="https://www.w3.org/TR/webxr/">WebXR Device API</a> specification is still in draft, although it has been partially implemented in Chrome 79. If you want to learn more about this technology, check out this <a href="https://developer.mozilla.org/en-US/docs/Web/API/WebXR_Device_API">MDN page</a>, as well as these cool <a href="https://immersive-web.github.io/webxr-samples/">examples</a>.</blockquote><p>One of the most popular library that brings Augmented Reality to the web is <a href="https://github.com/AR-js-org/AR.js">AR.js</a> (developed by Jerome Etienne). This library uses <a href="https://github.com/kig/JSARToolKit">JSARToolkit</a> for the image tracking, and can display 3D content with either <a href="https://threejs.org/">Three.js</a> or <a href="https://aframe.io/">A-Frame</a>.</p><p>It supports the following types of augmented reality:</p><ul><li><strong>Marker-based AR, </strong>which<strong> </strong>uses patterns like QR codes or other custom designs to trigger an augmented experience. When a marker in the physical world is recognized by a device’s camera, the digital content is placed on top of it.</li><li><strong>Image tracking, </strong>similar to marker-based, it makes it possible to use 2D images to display content, instead of the classic marker.</li><li><strong>Location-based AR, </strong>ties digital content and the experience it creates to a specific place. The objects are mapped out so that when a user’s location matches the predetermined spot it is displayed on the screen.</li></ul><h3>Use case example</h3><p>For the purpose of this article, I’ve created a simple project with AR.js and <a href="https://vuejs.org/">Vue</a> JS, to showcase some AR capabilities in a real-world use case.</p><p>In <a href="https://medium.com/tuimm"><strong>Tui Musement</strong></a>, we provide our customers with thousands of experiences and excursions everywhere in the world. When users buy an experience on one of our platforms, they receive a ticket, which they may need to print to show it to the provider of that experience.</p><p>One possible way to improve our customer experience, and to add a “wow factor” to our products, would be to use Augmented Reality to show some dynamic information to our customer, directly on their ticket. The possibilities are endless, but to keep things simple I’ve chosen to show some simple hard-coded information to tell our customer how to reach the meeting point. Obviously in a real-world we would have pulled this data from an API, but this is just a starting point. You can easily expand this project with all the functionality you need.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/938/1*sfjuN7EaL6dBerI7c7S3dg.png" /><figcaption>Icons by Icons8</figcaption></figure><p>The logic is very simple. We use a QR Code to redirect our customer to our WebAR app (you can also use a custom-generated QR code to pass some data to the app, like an order reference number, through a query string). The custom marker is then used to render the content on the voucher.</p><h4>Project Setup</h4><p>First of all, let’s create a new project using vue-cli:</p><pre>vue create my-awesome-ar-project</pre><p>Next, we need to install AR.js. Unfortunately, there’s no npm package available at the moment, so we need to import the library directly in our index.html. You can find AR.js and A-Frame build files <a href="https://github.com/GabriMcNab/augmented-reality-experiment/tree/master/public/js">here</a>, or you can just use the CDN, as shown on the <a href="https://ar-js-org.github.io/AR.js-Docs/#import-the-library">documentation</a>.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/a6981baec75aa7a22a506e467de03391/href">https://medium.com/media/a6981baec75aa7a22a506e467de03391/href</a></iframe><p>The A-Frame version of AR.js uses custom HTML elements to define the scene and all of it’s components. We need to tell Vue to ignore those elements, otherwise we will get a bunch of warnings printed in the console. Here we are telling Vue to ignore all elements that start with “a-” (a-scene, a-entity, etc).</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/12e1010a2586528d59f4847aa36d30e6/href">https://medium.com/media/12e1010a2586528d59f4847aa36d30e6/href</a></iframe><h4>Basic Scene</h4><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/4840e240b8dad388e087d6daa66add24/href">https://medium.com/media/4840e240b8dad388e087d6daa66add24/href</a></iframe><p>In App.vue, we use &lt;a-marker&gt; to enable the marker tracking and render a 3D model (defined by the &lt;a-entity&gt; component) once the marker is found by the device’s camera. By default, you can use two marker presets, “hiro” and “kanji”, which look like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/967/1*Y7Uwni793IvLHi2FjY3Mcw.png" /></figure><p>For this project, I wanted to use a custom marker with our Tui logo. To make it, I’ve used this <a href="https://jeromeetienne.github.io/AR.js/three.js/examples/marker-training/examples/generator.html">online tool</a>. Here’s what I came up with:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*wa94DsTc9aqoYlxjclWpmA.png" /></figure><p>Once downloaded, you will find a .patt file. This is what we need to include in our project (Note: you will also need to specify type=”pattern” in order to make it work)</p><pre>&lt;a-marker <em>type</em>=”pattern” <em>url</em>=”path-to-patt-file&quot;&gt;</pre><h4>Display Content in the scene</h4><p>At this point I’m ready to render some content in the scene. First thing first, I would like to show some information to tell our customer how to reach the meeting point, as well as the distance and the travel time.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/5a793fcc85386280995930fe87071582/href">https://medium.com/media/5a793fcc85386280995930fe87071582/href</a></iframe><p>Obviously this data is fake, but you can see how easy it could be at this point to implement an API call in order to fetch some dynamic data.</p><p>This is the component I’ve used to display the information:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/18abf97d94e12b88d22bc5748691946c/href">https://medium.com/media/18abf97d94e12b88d22bc5748691946c/href</a></iframe><p>Note: the units of measurement for dimensions (width, height) and position are directly proportional to the dimension of the marker in the real world. For example, if you print a marker that is 10x10cm, and you add a component like this one:</p><pre>&lt;a-plane width=”2&quot; height=”1&quot; /&gt;</pre><p>You will obtain a rectangle that (in real-world dimensions) will be 20x10cm. This makes it pretty simple for us to understand how to position the elements on the page, since we know how big the marker will be once we print the ticket.</p><p>The last component I’d like to implement for this project, is a button that allows our user to open the directions directly on the Google Maps app. Thanks to the <a href="https://aframe.io/docs/1.2.0/components/raycaster.html#sidebar">raycaster component</a> of A-Frame, we can check for user interactions (like click and gestures) with all the objects displayed on the screen. This allows us to attach some event handler to our objects:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/e32ad5a5eebafb659574f911d72b8cd5/href">https://medium.com/media/e32ad5a5eebafb659574f911d72b8cd5/href</a></iframe><p>Here’s a demo of the final result:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FKsPqnS00zp8%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DKsPqnS00zp8&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FKsPqnS00zp8%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/c0a2b42a1968178832c234fe3e1a12e3/href">https://medium.com/media/c0a2b42a1968178832c234fe3e1a12e3/href</a></iframe><blockquote>Note: On some devices you could notice some issues with the click event being off-center. This is an open issue, you can check this <a href="https://github.com/AR-js-org/AR.js/issues/40">link</a> if you want to contribute.</blockquote><h3>Conclusion</h3><p>I’ve had a lot of fun researching this technology for this article. I think it has a lot of potential and you should definitely check it out. As <strong>Tim Cook</strong>, Apple’s CEO, said:</p><blockquote>I do think that a significant portion of the population of developed countries, and eventually all countries, will have AR experiences every day, almost like eating three meals a day. It will become that much a part of you.</blockquote><p>If you want to learn more about AR.js, check out this <a href="https://medium.com/swlh/ar-js-the-simplest-way-to-get-cross-browser-augmented-reality-on-the-web-10cbc721debc">great article</a> by Nicolò Carpignoli.</p><p><a href="https://github.com/GabriMcNab/augmented-reality-experiment">Here</a>, you can find the repo I’ve created for this article, as well as a link to the live example.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=350bef8a44f2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/elevate-your-web-app-with-augmented-reality-350bef8a44f2">Elevate Your Web App With Augmented Reality</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The future of digital advertising in the cookieless world and the importance of 1st party data]]></title>
            <link>https://medium.com/tuimm/the-future-of-digital-advertising-in-the-cookieless-world-and-the-importance-of-1st-party-data-f7100718a21c?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/f7100718a21c</guid>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[cookies]]></category>
            <category><![CDATA[content-mode]]></category>
            <dc:creator><![CDATA[Pol Badia]]></dc:creator>
            <pubDate>Thu, 24 Feb 2022 14:32:56 GMT</pubDate>
            <atom:updated>2022-02-24T14:32:56.488Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>The future of digital advertising in the cookieless world and the importance of 1st party data</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zrZPLSZDh6KTd-yrqdtgjg.jpeg" /></figure><p>We all love cookies, sometimes we can find them with chocolate, butter or even with ginger. From a digital point of view, cookies do not have such a good reputation. Cookies are small files of data that a web server generates and sends to a browser which, afterwards, inform websites about the user, allowing the websites to personalize the user experience. We can differentiate 2 types of cookies:</p><p><strong>1st party cookies</strong>: Are created by the website a user visits and collects information about that user, such as username, passwords, language preference, payment method and items in a shopping cart. Thanks to this, the website does not have to ask the user to enter his information every time he returns, it can also make product suggestions based on his preferences and offer a better browsing experience.</p><p><strong>3rd Party cookies:</strong> Are created by a domain other than the one the user is visiting, and it is usually related with digital advertising. Main functions of these cookies are cross-site targeting, retargeting, ad serving or reach and frequency measurement.</p><p>The use of cookies has been crucial for digital marketers during these past years. We use them to improve user experience, to collect data, to track what visitors are doing on our website, etc. Cookies help us to target the perfect advertisement to the right audience and personalize user experience across all digital channels. It also plays a key role in assigning conversions to the channel that drove them which is essential for good campaign optimization. Nowadays, competing in the open market for a position in the search page results is difficult but if you cannot track your performance, it is almost impossible to make it work.</p><p>3rd party cookies will “soon” be <strong>removed</strong> from the digital marketing landscape. At least, this is the plan. Some of Google’s main competitors, Safari and Mozilla Firefox, have already removed them and Google was planning to do so sometime this year. However, they have postponed it till 2023. Seems that they need more time to adapt to all these advertising ecosystems to not harm themselves too much. This makes total sense considering that a big percentage of Google’s revenue comes from advertising. Vinay Goel (privacy engineering director of Chrome), <a href="https://blog.google/products/chrome/updated-timeline-privacy-sandbox-milestones/">confirms it</a>: “While there’s considerable progress with this initiative, it’s become clear that <strong>more time is needed</strong> across the ecosystem to get this right.”</p><p>As an alternative to what we advertisers would call a “cookieless world”, what Google proposes is the <a href="https://privacysandbox.com/">Privacy <strong>Sandbox</strong></a><strong>.</strong> Privacy Sandbox is a group of standards that ensure protection of the privacy of all users, but at the same time guarantees the personalization of their ads. One of its more interesting ideas is the “<a href="https://privacysandbox.com/proposals/floc#:~:text=Federated%20Learning%20of%20Cohorts%20(FLoC)%20is%20a%20new%20way%20for,tracking%20individuals%20across%20the%20web."><strong>Floc</strong></a>” (Federated Learning of Cohorts). What it does is collect information about browsing habits to assign each user into a specific group/flock considering their browsing histories and habits. Doing this we are not using the users as individuals that we can identify but as part of a group.</p><p>With all these changes going on, it is still uncertain how it will affect our marketing strategies. Therefore <strong>1st party cookies</strong> become even more important. With first-party cookies we discover what users are doing while visiting our website. We are talking about returning users, which makes them extremely valuable due to the fact they are likely to know and like our brand. Therefore, we can use all this information to create relevant marketing campaigns for every user.</p><p>Since October 2019, Europe’s top court has forced all the websites to have explicit consent of the users before storing or accessing non-essential<strong> 1st party cookies</strong>, such as tracking cookies for advertising or analytical purposes. If we have a look back, it has been a long path from where we started to where we are now. <em>(Data from </em><a href="https://gdpr.eu/what-is-gdpr/"><em>gdpr.eu</em></a><em>).</em></p><ul><li>In 1994 the first banner ad appeared online.</li><li>In 1995 it passed the European Data Protection Directive, establishing minimum data privacy and security standards, upon which each member state based its own implementing law.</li><li>In 2012 the proposal for the GDPR was released.</li><li>In 2016 the GDPR entered into force in 2016 after passing the European Parliament.</li><li>In 2018 all organizations were required to be compliant.</li><li>In 2019 Organizations are forced to have <strong>explicit consent</strong> before storing non-essential 1st party cookies.</li></ul><p>Seems like a lot of legal and technical information that should not affect us at all, you would think. However, have you noticed that all web browsing is constantly interrupted by huge pop-up banners that ask your explicit permission for tracking your activities on the websites? This is how the 2019 resolution directly and indirectly affects all of us, as users decide the level of privacy that they want to concede to each website. This of course is a huge accomplishment to restoring our online privacy, but at the same time, it makes it a lot more difficult for us marketers to analyze the performance of marketing channels when users decline consent because we lose track of all their actions on our web. While we are counting the cost and clicks from google ads, we still may lose crucial information on website conversions. Therefore, we have less information and relevant signals to nurture machine learning algorithms for bidding and optimization.</p><p>To solve that problem, google has developed a tool called “google consent mode.” This allows you to adjust how your Google tags (Google Tags are tags that we implement in the website source code, whose objective is to send information to the measurement platforms) behave based on the consent status of your users. The data will be observed as normal if the user gives consent. When users do not give their explicit consent for tracking and storing their information, the relevant Google tags will adjust their functions and model user behavior, which will then show the most accurate data in your Google Ads account adding the extra conversions not tracked to enable better reporting and optimization for marketing purposes.</p><p>In <a href="https://careers.tuigroup.com/careers/">Tui Musement</a>, with the support of our Google account managers, we were able to implement <a href="https://support.google.com/analytics/answer/9976101?hl=en">Google Consent Mode</a> in April 2021 with the help of our IT team. They adjusted and updated tags behavior in Google Tag Manager adding javascript codes, so that the users of Musement websites were split into two groups: those who accepted cookies and those who did not. Extra parameters were then also added to the tags, so that they could report aggregated data to our tracking platforms as opposed to no data at all (without using the feature) of people who opted out of cookies.</p><p>By setting up Consent Mode in our accounts we achieved an <strong>increase in measured conversions</strong> of 5,3% in IT, +3,9% in DE, +3,4% in FR, +25% in ES, +9,1% in GB, resulting in a decrease of the lost data impact on *smart bidding algorithms. Our campaigns were able to receive more relevant signals from Google and website conversions and the impact of each campaign was much more visible. Being this quite a substantial uplift in campaigns metrics reporting, Google also drew up a <a href="https://www.thinkwithgoogle.com/intl/it-it/futuro-del-marketing/digital-transformation/privacy-strategie-misurazione/">case study</a> to showcase our success and inspire more companies like Musement to achieve better results in their campaign performance.</p><p><strong><em>*</em></strong><em>Smart bidding are automated strategies that uses machine learning to optimize the advertising bidding considering historical data and then optimizing it accordingly in each auction.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f7100718a21c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/the-future-of-digital-advertising-in-the-cookieless-world-and-the-importance-of-1st-party-data-f7100718a21c">The future of digital advertising in the cookieless world and the importance of 1st party data</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Gearing Up for the Next Stage of Machine Learning]]></title>
            <link>https://medium.com/tuimm/gearing-up-for-the-next-stage-of-machine-learning-96c9d94b8ac8?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/96c9d94b8ac8</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[digital-transformation]]></category>
            <dc:creator><![CDATA[Carola Gerwig]]></dc:creator>
            <pubDate>Mon, 14 Feb 2022 10:16:30 GMT</pubDate>
            <atom:updated>2022-02-15T15:19:20.555Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BDTYZHsvpkgqQ9ePmtwpmw.jpeg" /><figcaption>Data Science Journey: What we do &amp; how we do it</figcaption></figure><p>At <a href="https://careers.tuigroup.com/">TUI </a>we use data science in every facet of our business. We have dedicated data science teams in each of our business units. These teams have implemented <a href="https://en.wikipedia.org/wiki/Machine_learning">machine learning</a> algorithms in a variety of fields like better inventory planning and by dynamically adjusting our offerings to the customers.</p><p>We use bottom up processes guided by our data scientists to align and innovate on data science life-cycle processes across the company. Besides building new ML based products, we are gearing up for the next stage of TUI’s machine learning journey. Our next objectives can be categorized in terms of infrastructure and knowledge development. In this blog post we briefly describe these objectives.</p><p><strong>Infrastructure</strong></p><p>A good infrastructure should permit easy debugging, safe rollback, minute process monitoring, and reliable scaling for a large number of users. Moreover, it should readily accommodate to a wide-variety of architectures. Its goal is to enable a quick and seamless transition from development to deployment of models.</p><p>As our machine learning products developed, each team put in place infrastructure that enabled TUI to tackle problems of greater complexity that required more computation and other resources. Since we give maximum autonomy to our teams, each team has developed infrastructure suitable to the specific requirements of their own projects. These independent developments have led to redundancies as different teams have separately developed infrastructures for similar needs. Such redundancies are not only inefficient, but they also prevent a comprehensive understanding that can only be gained by solving similar problems in different contexts.</p><p>We explored many end-to-end infrastructure solutions on the market. While useful for initial iterations, they didn’t possess the rigor and flexibility to serve our advanced modeling requirements. So, we developed TUI Analytics Platform (TAP). TAP is an integrated solution that combines <a href="https://kubernetes.io/">Kubernetes</a>, <a href="https://about.gitlab.com/">Gitlab</a>, <a href="https://airflow.apache.org/">Airflow</a>, <a href="https://mlflow.org/">MLFlow</a>, <a href="https://www.datadoghq.com/">Datadog</a>, <a href="https://www.snowflake.com/">Snowflake</a>, <a href="https://aws.amazon.com/s3/?nc1=h_ls">S3</a> based data products, and other tools to deliver a flexible and comprehensive solution. TAP is being continuously developed to add new capabilities and other improvements based on feedback of users such as data scientists and ML engineers.</p><p>Believing in tight coupling between model development and model industrialization, TAP requires data scientists to carefully reason about their requirements. While it doesn’t require them to be fully versed in the intricacies of our platform, it does encourage them to develop an understanding of the tools which greatly helps to structure the implementations well from the start. As a matter of fact, there can be a learning curve for some data scientists. One of our major focuses has been to motivate other data scientists to consistently use this platform that has proven its value in production for many projects.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/605/1*QMEE2xQNX1vdHm-87gkoyA.png" /><figcaption>Outline of the TUI Analytics Platform for Data Science at TUI</figcaption></figure><p><strong>Knowledge Development</strong></p><p>This brings us to the human dimension of our objectives. We encourage our data scientists to write production ready code from the get-go. Over the years, different teams have adopted different sets of development practices.</p><p>We have observed that a certain set of processes and guidelines inspired by proven software development practices provides the best trade-off between the speed of experimentation and the ease and reliability of industrialization of models. To promote these practices we have developed a comprehensive template that helps less experienced data scientists in software development to start working with TAP while serving as canonical guidelines for more experienced colleagues.</p><p>In addition, more than any other field of IT, modeling benefits from perspective of people with diverse professional skills and experience. So we consider it particularly important to enhance the cooperation between members of different teams that are now distributed all over the globe. At TUI we have regular events including informal company-wide <a href="https://en.wikipedia.org/wiki/Stand-up_meeting">stand-up meetings</a> to share our experiences, tutorials and workshops about different data science and MLOps topics, and deep dive sessions about ongoing ML work or research topics. This has helped maintain a high communal spirit among our data scientists.</p><p><strong>Looking Ahead</strong></p><p>TUI has rapidly transformed over the last two years to successfully surmount challenges posed by the pandemic. We have used this challenge to successfully innovate on data science processes. By unifying infrastructure and ways of working we can save development resources and benefit from the expertise across the organisation while empowering our globally distributed teams to respond quickly to their local teams or market conditions. We intend to continue on this path. Among the tools we used to push forward is internal competitions, we will describe one in an upcoming blog post.</p><p>Thanks to my co-author <a href="https://medium.com/@jasdeep.singh_60041">Jasdeep Singh</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=96c9d94b8ac8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/gearing-up-for-the-next-stage-of-machine-learning-96c9d94b8ac8">Gearing Up for the Next Stage of Machine Learning</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A day in the life of a TUI Musement Software Engineer]]></title>
            <link>https://medium.com/tuimm/a-day-in-the-life-of-a-tui-musement-software-engineer-89c337923740?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/89c337923740</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[developer-stories]]></category>
            <category><![CDATA[daily-life]]></category>
            <dc:creator><![CDATA[Gionatan Lombardi]]></dc:creator>
            <pubDate>Fri, 03 Dec 2021 14:56:22 GMT</pubDate>
            <atom:updated>2021-12-03T17:32:15.319Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SlO-YAGOsNfvRb1w7gpiRQ.jpeg" /><figcaption>Photo by <a href="https://www.pexels.com/@anikinearthwalker?utm_content=attributionCopyText&amp;utm_medium=referral&amp;utm_source=pexels">Dimitry Anikin</a> from <a href="https://www.pexels.com/photo/colorful-old-fashioned-clock-in-city-4861258/?utm_content=attributionCopyText&amp;utm_medium=referral&amp;utm_source=pexels">Pexels</a></figcaption></figure><p>In this article, I will present a snapshot of daily life as a Software Engineer at TUI Musement and give some insights into the way we work.</p><h3>9 AM</h3><p>Time to start! Let’s grab a coffee and, when I’m in the office, have a quick chat with colleagues: perhaps there’s even some delicious cake if I’m lucky!</p><p>In TUI Musement, the Engineering Teams work with the Agile methodology using the Scrum or Kanban framework, so for the first hour (give or take) it’s preparation time: we prepare for the stand-up, update tasks in the Jira platform, check emails, and organize our daily work.</p><p>During the week we’ll also attend the usual Agile Ceremonies like Planning, Refinement, and Sprint Retrospective.</p><h3>10 AM</h3><p>Most of the time we work remotely and as many colleagues are located in different countries, all of the meetings are video calls. However, when we’re in the office I love to sit in the same room as my teammates and attend the meeting together.</p><p>It’s time to update everyone about the work I completed yesterday and what I’m going to do today. We usually report blockers and issues we run into and re-align on technical decisions. All developers attend, together with the Product Manager, the Product Designer, and the Scrum Master.</p><p>In TUI Musement we’re organized into Agile Domain Teams: each team is responsible for a specific area of the business, like payments, orders, products, content, integrations, and many others.</p><p>We work on different platforms: B2C websites (like <a href="https://www.musement.com/">www.musement.com</a> and <a href="https://www.gotui.com/">www.gotui.com</a>), B2B solutions for suppliers, partners, and affiliates, and internal tools for data management. The teams share the ownership of these applications, implementing solutions related to their specific domain.</p><p>Historically our data source and API provider is a PHP monolith reading and writing from a MySQL Database but recently we’ve started to extract functionalities into separate microservices written in Golang, Java, and .NET as it is more tailored to the needs and separations required by the domains.</p><p>In addition to that, NodeJS, Golang, or PHP API Proxies serve specific web applications following the <a href="https://medium.com/frontend-at-scale/frontend-architectural-patterns-backend-for-frontend-29679aba886c">Backend For Frontend architecture</a>. We use them to merge and select data from different APIs so the frontends will receive only those strictly needed.</p><p>Most of our frontends are written in <a href="https://vuejs.org/">VueJS</a> and more specifically, with the <a href="https://nuxtjs.org/">NuxtJs</a> framework. Since the codebases are shared between teams, we required a fast and consistent way to build them. NuxtJS is serving this purpose very well since its structure is simple and strict so we know we’ll find the same patterns across all our platforms!</p><p>Also, on the Frontend, we use the so-called <a href="https://medium.com/tuimm/vue-shared-components-2c40973c32c5">Shared Components</a>: VueJS components, developed in a separate repository, which can be configured and imported as NPM packages inside the NuxtJS applications. Thanks to this, we can plug and play shared functionalities (search, product pages, checkout, and payment flow) across all platforms without rewriting them all the time.</p><h3>10:15 AM</h3><p>Stop talking, let us get some work done!</p><p>During the Daily Scrum, our Product Manager pointed out a bug that urgently needs to be addressed. It’s not clear if the issue relies on the backend or in the web application.</p><p>I think it’s a good idea to reach out to one of my frontend colleagues on Slack and see if she has some time to do a pair programming session with me for the analysis. Luckily, she has already started to investigate the issue and can provide me with the <a href="https://en.wikipedia.org/wiki/CURL">cURL</a> that is replicating the issue.</p><p>It seems that the controller of the endpoint, inside our new microservice, is not handling an edge case. Before implementing the fix, I’m going to write the unit test for that case and, as expected it fails. Now I can work on the controller and finally the test is passing: looking at the green ✅ is always a big satisfaction!</p><p>The next step will be to restart the service inside the docker container and locally check if the cURL is now returning the expected response. It’s working! 🎉. Now I can commit the fix and open a Merge Request in the Gitlab repo, adding all my backend colleagues within the team as reviewers. For the Merge Request, we have a template that helps us to explain the changes in detail, highlighting the WHAT, the WHY, and HOW to TEST them.</p><p>In the description, I will assume that the reader has a low context about the project and the code base, so I add a lot of extra information.</p><p>After my commits, a pipeline will start, running the unit tests and deploying a feature branch to an ephemeral environment. I’ll add the cURL in the Jira ticket so the fix can also be validated by the testers before releasing it to production.</p><h3>12:00 AM</h3><p>On Slack, we have a channel in which we report the new MR we opened. I add mine but I also see that I need to review some others opened by my colleagues.</p><p>I spotted one on an API Proxy I was working on during the previous Sprint; I have a lot of knowledge on that code-base: maybe I can give some good suggestions and finish the review before lunch.</p><h3>1 PM</h3><p>It’s lunchtime! Yesterday I went to the office and we decided to go to our favorite <a href="https://www.facebook.com/RistoranteBajaSardinjaMilano/">seafood restaurant</a>. They prepare delicious dishes of Sardinian cuisine and it’s always a big pleasure for us all to go there.</p><p>Today it’s different, I’m at home and luckily, it’s a sunny day: the perfect weather for a short jog around the park nearby my home and an easy-fresh meal before reopening the laptop. This will also help to re-establish the balance after the generous meal we had yesterday: they always treat us very well 😋.</p><h3>2 PM</h3><p>Today I have a 1:1 with my manager. In this weekly appointment, I can share updates, learning, and achievements and she will give (and ask for) feedback. Above all, it’s a space to discuss skill development and career growth.</p><p>I usually prepare the topics I want to discuss in advance, to be sure we cover all the points that are important to me.</p><h3>2:45 PM</h3><p>If I don’t have other meetings in the afternoon, I can focus on a new feature I started to develop a few days ago. I want to introduce an external library in my application, but I’m not sure it’s a good idea so I reach out to a colleague in the Performance Team. The Performance Team is a cross-domain team, focused on performance, shared standards, and best practices. She proposes a simpler alternative already used by another team, that is perfect for my use case. I’m able to complete the implementation and open another Merge Request.</p><h3>5 PM</h3><p>It’s time to write down an idea I had and want to explore. I’ve often heard about <a href="https://graphql.org/">GraphQL</a>, but I don’t know it well and I’ve never had the time to research it or use it in a real-world project. My team is using <a href="https://en.wikipedia.org/wiki/Representational_state_transfer">RESTful</a> APIs, and we never really considered it as an alternative. I would like to study it, to understand if we could benefit from it. This topic seems like a perfect candidate for Tech Friday!</p><p>Tech Friday is one of the things I love the most about working in TUI Musement. We can dedicate a day every week (Friday, as you can imagine) to develop our skills, both hard and soft. I can’t imagine a better way to conclude the week. Software engineers have the chance to decide how to spend their time and the projects they will work on. We can choose to study new technologies, experiment, read a book or improve the existing software, focusing on aspects like performance or monitoring. We can do it both individually or in collaboration with one or more teammates. At the end of the project, we’ll share the result with the colleagues, so that they can benefit from it too.</p><p>The outcome can be a<a href="https://en.wikipedia.org/wiki/Proof_of_concept"> </a>proof of concept, in which we demonstrate the feasibility of an idea; an internal tech meet-up; an<a href="https://adr.github.io/"> architectural decision record</a> to document a software design choice; a<a href="https://www.atlassian.com/software/confluence"> Confluence</a> page to be shared with the company.</p><p>Sometimes we write articles describing a project which started during Tech Friday. You can find them on <a href="https://medium.com/tuimm">our tech blog</a>. Besides the already mentioned shared components approach, a good example is <a href="https://medium.com/tuimm/measure-performance-web-app-with-lighthouse-ci-in-a-gitlab-pipeline-dd292842e40d">how to measure the performance of a web application with Lighthouse CI in a GitLab CI pipeline</a>. Or the article describing <a href="https://medium.com/tuimm/implement-a-ui-library-with-web-components-de0bffcf39ff">how to create a UI library with Web Components.</a></p><p>Today I want to formalize my proposal about GraphQL. I will spend some Fridays trying to analyze the pros and cons, and together with a colleague, I will implement a PoC, to make sure that it fits all our requirements. I send the plan to my engineering manager, who promptly approves it. Tomorrow will be an interesting day.</p><h3>6 PM</h3><p>It’s time to turn off the pc. If I’m working from home, I can spend some quality time with my family and my friends. If I’m at the office, it’s time to think about the <em>aperitivo </em>with my colleagues.</p><h3>Kudos</h3><p>A big thank you to <a href="https://medium.com/u/afff1cdfeb06">Marco Ziliani</a> who helped me write this article.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=89c337923740" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/a-day-in-the-life-of-a-tui-musement-software-engineer-89c337923740">A day in the life of a TUI Musement Software Engineer</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[.NET Code coverage on GitLab]]></title>
            <link>https://medium.com/tuimm/net-code-coverage-on-gitlab-3e35f08d0f44?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/3e35f08d0f44</guid>
            <category><![CDATA[dotnet]]></category>
            <category><![CDATA[gitlab-ci]]></category>
            <category><![CDATA[code-coverage-tools]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Leonardo Gasparini]]></dc:creator>
            <pubDate>Thu, 04 Nov 2021 11:20:20 GMT</pubDate>
            <atom:updated>2021-11-04T11:20:20.198Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LlzLHDOo1j0sOpN05jmWoA.jpeg" /></figure><p>Testing our software is a crucial development stage to maintain a high standard of quality. To reduce the amount of bugs being released into the production environment, code coverage, along with other tests, can be used as a helpful measurement index.</p><p>Code coverage is a measure that describes how much of a software source code is executed when a particular test suite runs. High test coverage means that the code has a lower chance of containing undetected bugs compared to a program with lower outcomes.</p><p>There are many different metrics that can be used to calculate test coverage; some of the most basic are the percentage of lines and branches called during execution of the test suite.</p><p>To monitor it, GitLab, through its CI/CD pipelines, offers the possibility to run your tests, parse the coverage percentage from the <em>stdout</em> through a regular expression and show it on merge requests.</p><p>Unfortunately, this isn’t the case with .NET projects, since there is no built-in support from Gitlab to .NET, but in <a href="https://www.linkedin.com/company/musement/jobs/">TUI Musement</a> we care about quality, so we found a solution!</p><p><strong>NOTE</strong>: you <strong>should not rely on coverage percentage alone</strong>, tests must be properly written to cover any corner case your code could face!</p><h3>Get started</h3><p>In order to produce the necessary <a href="https://docs.gitlab.com/ee/ci/pipelines/job_artifacts.html">artifacts</a>, we add 2 packages to the test project: <a href="https://github.com/coverlet-coverage/coverlet"><strong>Coverlet</strong></a><strong>, </strong>a cross platform framework with support for line, branch and method coverage<strong>, </strong>and <a href="https://github.com/spekt/junit.testlogger"><strong>Junit</strong></a><strong>, </strong>a tool that allows to output the XML format GitLab needs.</p><pre>dotnet add package coverlet.collector<br>dotnet add package JunitXml.TestLogger</pre><p>You can find a sample project <a href="https://gitlab.com/lg.musement/coverage-sample">here</a>.</p><h3>Run tests</h3><p>First, we need to make sure to run our tests properly and collect the coverage data, to do that we use the following command paying particular attention to the —-collect and —-logger options:</p><pre>dotnet test -c Debug<br>            -r ./cobertura<br>            --collect:&quot;XPlat Code Coverage&quot;<br>            --test-adapter-path:.<br>            --logger:&quot;junit;LogFilePath=./junit/junit-test-result.xml;MethodFormat=Class;FailureBodyFormat=Verbose&quot;</pre><p>This produces the following files containing all the needed information.</p><pre>cobertura/<br>  {auto generated hash}/<br>    coverage.cobertura.xml<br>...<br>tests/<br>  Weather.Tests/<br>    junit/<br>       junit-test-result.xml</pre><p>It’s time to automate this process! Add a job to .gitlab-ci.yml in order to get the work done for us and use the generated artifacts automatically every time the pipeline is triggered according to the defined rules. In this case, to keep it simple, they have been omitted.</p><pre>test-with-coverage:<br>  image: mcr.microsoft.com/dotnet/sdk:5.0-alpine<br>  stage: tests<br>  variables:<br>    CONFIGURATION: &quot;Debug&quot;<br>    COVERAGE_FLAG: &quot;XPlat Code Coverage&quot;<br>    LOGGER_FLAG: &quot;junit;LogFilePath=$CI_PROJECT_DIR/junit/junit-test-result.xml;MethodFormat=Class;FailureBodyFormat=Verbose&quot;<br>  script:<br>    - &#39;dotnet test<br>            -c $CONFIGURATION<br>            -r $CI_PROJECT_DIR/cobertura<br>            --collect:&quot;$COVERAGE_FLAG&quot;<br>            --test-adapter-path:.<br>            --logger:&quot;$LOGGER_FLAG&quot;&#39;<br>  artifacts:<br>    paths:<br>      - $CI_PROJECT_DIR/cobertura/*/coverage.cobertura.xml<br>      - $CI_PROJECT_DIR/junit/junit-test-result.xml<br>    reports:<br>      cobertura:<br>        - $CI_PROJECT_DIR/cobertura/*/coverage.cobertura.xml<br>      junit:<br>        - $CI_PROJECT_DIR/junit/junit-test-result.xml</pre><p>Using the mcr.microsoft.com/dotnet/sdk:X.X-alpine <a href="https://www.docker.com/">Docker</a> image to execute the dotnet test command, we’re telling GitLab to retrieve the XML files through the paths keyword (it’s important to use the wildcard * since the folder name is an auto generated hash) and use them as reports through the homonymous keyword.</p><p>This allows GitLab to show the covered lines directly into the merge request, but, since the dotnet test command doesn’t print it to <em>stdout</em>, the total coverage percentage is still missing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/980/1*Hb6XGC6zKzJZDKKiaQbafw.png" /></figure><h3>Retrieve the coverage percentage</h3><p>Since the test command doesn&#39;t produce any output concerning the total coverage percentage, we need to extract it from the newly created coverage.cobertura.xml report and print it to <em>stdout </em>using the following script:</p><pre><strong>#!/usr/bin/env sh</strong></pre><pre>REPORTS_DIR=&quot;<em>$</em>{1}&quot;</pre><pre><em>if [[ </em>$# -lt 1 <em>]]</em>; <em>then<br>    echo </em>&gt;&amp;2 &quot;You have to provide the directory containing your coverage reports.&quot;<br>    <em>exit </em>2<br><em>fi</em></pre><pre>total_coverage=0<br>count=0</pre><pre><em>for </em>i <em>in $(find </em>&quot;$REPORTS_DIR&quot; -name &#39;*.xml&#39;<em>)</em>;<br><em>do<br>    printf </em>&quot;Found coverage report: %s\n&quot; &quot;$i&quot;<br>    line_rate=&quot;<em>$(head </em>-n 2 <em>$</em>{i} | <em>sed </em>&#39;N;s/.*line-rate=&quot;\([^&quot; ]*\).*/\1/g&#39;<em>)</em>&quot;<br>    coverage=<em>$(echo </em>&quot;<em>$</em>{line_rate} * 100&quot; | <em>bc)</em></pre><pre><em>    printf </em>&quot;PARTIAL_COVERAGE: %2.2f\n&quot; &quot;$coverage&quot;</pre><pre>    count=<em>$</em>((count + 1))<br>    total_coverage=<em>$(echo </em>&quot;<em>$</em>{total_coverage} + <em>$</em>{coverage}&quot; | <em>bc)<br>done</em>;</pre><pre><em>printf </em>&quot;Found a total of %i report(s)\n&quot; &quot;$count&quot;<br><em>printf </em>&quot;TOTAL_COVERAGE=%2.2f\n&quot; &quot;<em>$(echo </em>&quot;<em>$</em>{total_coverage} / <em>$</em>{count}&quot; | <em>bc </em>-l<em>)</em>&quot;</pre><p>Update the test-with-coverage job to make it execute the script by adding these new lines under the script<em> </em>keyword:</p><pre>- chmod +x ./scripts/print-dotnet-coverage.sh<br>- ./scripts/print-dotnet-coverage.sh $CI_PROJECT_DIR/cobertura</pre><p>Define the regular expression that will parse the <em>stdout</em> to get the desired value by adding this new keyword to the job:</p><pre>coverage: /TOTAL_COVERAGE=(\d+.\d+)/</pre><p>This will finally allow us to see the coverage percentage directly into the merge request.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/834/1*zxadb_MIjM0teQBzIBFjaQ.png" /></figure><h3>Publish the report</h3><p>Now, in order to make the report easily available for consultation, we generate an html version using <a href="https://github.com/danielpalme/ReportGenerator">ReportGenerator</a> and publish it to a GitLab environment introducing the following job:</p><pre>deploy-coverage-report:<br>  image: mcr.microsoft.com/dotnet/sdk:5.0-alpine<br>  stage: deploy artifacts<br>  needs:<br>    - job: test-with-coverage<br>      artifacts: true<br>  variables:<br>    PUBLIC_URL: &quot;/-/$CI_PROJECT_NAME/$CI_PROJECT_NAME/-/jobs/$CI_JOB_ID/artifacts/public&quot;<br>  environment:<br>    name: Coverage/MyWeatherAPI<br>    url: &quot;https://$CI_PROJECT_ROOT_NAMESPACE.gitlab.io/-/$CI_PROJECT_NAME/-/jobs/$CI_JOB_ID/artifacts/public/index.html&quot;<br>  script:<br>    - dotnet tool install --tool-path . dotnet-reportgenerator-globaltool<br>    - mkdir -p public<br>    - ./reportgenerator &quot;-reports:$CI_PROJECT_DIR/cobertura/*/coverage.cobertura.xml&quot; &quot;-targetdir:public&quot; &quot;-reporttypes:Html&quot;<br>  artifacts:<br>    paths:<br>      - public</pre><p>In this job we’re inheriting the artifacts from test-with-coverage, using the aforementioned tool to generate the html report and telling GitLab to publish the result to an environment named MyWeatherApi under the Coverage folder.</p><p>After the next successful run, you can go to <strong>Deployments &gt; Environments, </strong>select the right one and click <strong>View deployment</strong> to reach your up-to-date coverage report and explore it. <br>A link will also be directly available into every interested merge request.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/979/1*itTsGG8003LmkAp9498d6g.png" /></figure><p>Now .NET projects can finally benefit from all GitLab code coverage features and, thanks to the deployed environment, you can analyze the report whenever you want to make sure all values are stable or decide if improvements are needed.</p><p><strong><em>Useful links:</em></strong></p><p><a href="https://docs.gitlab.com/ee/ci/">https://docs.gitlab.com/ee/ci</a><br><a href="https://github.com/coverlet-coverage/coverlet">https://github.com/coverlet-coverage/coverlet</a><br><a href="https://github.com/spekt/junit.testlogger">https://github.com/spekt/junit.testlogger</a><br><a href="https://github.com/danielpalme/ReportGenerator">https://github.com/danielpalme/ReportGenerator</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3e35f08d0f44" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/net-code-coverage-on-gitlab-3e35f08d0f44">.NET Code coverage on GitLab</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to use Apache Kafka without Java client?]]></title>
            <link>https://medium.com/tuimm/how-to-use-apache-kafka-without-java-client-555154096540?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/555154096540</guid>
            <category><![CDATA[software-architecture]]></category>
            <dc:creator><![CDATA[Gabriel Frontera Gayá]]></dc:creator>
            <pubDate>Wed, 20 Oct 2021 17:07:07 GMT</pubDate>
            <atom:updated>2021-10-20T17:07:07.360Z</atom:updated>
            <content:encoded><![CDATA[<h3>Introduction</h3><p>Apache Kafka was created with a client API only for Scala and Java. After that initial period, the Kafka Client API has been developed <a href="https://cwiki.apache.org/confluence/display/KAFKA/Clients">for many other programming languages</a>.</p><p>So, what if you want to use a development language that is unavailable? Or what if you want to access from an api gateway to Kafka?</p><p>In TUI Musement, we experienced the following scenarios:</p><ul><li>Use Kafka from SAPFI (SAP Financial Accounting, developed in ABAP)</li><li>Use Kafka from Apigee (the api gateway we are using)</li></ul><p>The solution we applied was to use Confluent Kafka Rest Proxy.</p><h3>What is Confluent Kafka Rest Proxy?</h3><p>Confluent Kafka Rest Proxy is an open source package developed by Confluent that allows users to access Kafka trough a RESTful API.</p><p>The API supports many interactions with your cluster, including message production and consumption, access to cluster metadata as well as management operations.</p><p>We can use it with binary, json or avro message. It also integrates well with the Confluent Schema Registry.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/675/1*4KL5ZZRQYT8Kq-ndYv0JPw.png" /></figure><p>It is available in <a href="https://docs.confluent.io/1.0/installation.html#installation">various format</a> with easy installation.</p><h3>How does the Confluent Kafka Rest Proxy work?</h3><p><strong>HTTP wrapper of Java libraries</strong></p><p>It simply wraps the existing libraries provided with the Apache Kafka open source project. This includes not only the producer and consumer you would expect, but also access to cluster metadata and admin operations. Therefore, it depends on specific versions of the Kafka libraries and the Confluent Platform packaging ensures the REST Proxy uses a compatible version.</p><p><strong>JSON with flexible embedded data</strong></p><p>Their requests need to include embedded data, the serialized key and value data that Kafka deals with. To make this flexible, confluent use <a href="https://www.confluent.io/docs/current/kafka-rest/docs/api.html#content-types">vendor specific content types</a> in Content-Type and Accept headers to make the format of the data explicit.</p><p><strong>Stateful consumers</strong></p><p>Consumers are stateful and tied to a particular proxy instance.</p><p><strong>Distributed and load balanced</strong></p><p>Despite consumers being tied to the proxy instance that created them, the entire REST Proxy API is designed to be accessed via any load balancing mechanism (e.g. discovery services, round robin DNS, or software load balancing proxy).</p><h3>Examples</h3><p><strong>Publish a message</strong></p><p>To publish a message you just need to send a json message with the following structure.</p><p>NOTE: you can publish one or various records at the same request (note that records field is a list).</p><p><em>Request</em></p><pre>curl --location --request POST &#39;<a href="https://kafka-rest-cmn.test.tui-dx.com/topics/suppliers&#39;">https://kafka-rest-cmn.test.tui-dx.com/topics/suppliers&#39;</a> \<br>--header &#39;Content-Type: application/vnd.kafka.json.v2+json&#39; \<br>--data-raw &#39;{<br>  &quot;records&quot;: [<br>    {<br>      &quot;key&quot;: &quot;1&quot;,<br>      &quot;value&quot;: {<br>        &quot;id&quot;: &quot;1&quot;,<br>        &quot;customer_code&quot;: &quot;1&quot;,<br>        &quot;telephone&quot;: &quot;888582158&quot;,<br>        &quot;email&quot;: &quot;<a href="/service/mailto:supplier1@test.com">supplier1@test.com</a>&quot;,<br>        &quot;language&quot;: &quot;EN&quot;        <br>      }<br>    }<br>  ]<br>}&#39;</pre><p><em>Response</em></p><pre>{<br>  &quot;offsets&quot;: [<br>    {<br>      &quot;partition&quot;: 0,<br>      &quot;offset&quot;: 3,<br>      &quot;error_code&quot;: null,<br>      &quot;error&quot;: null<br>    }<br>  ],<br>  &quot;key_schema_id&quot;: null,<br>  &quot;value_schema_id&quot;: null<br>}</pre><p><strong>Consume published messages</strong></p><p><em>Create a consumer</em></p><p>You need to create a consumer in a consumer group</p><p><em>Request</em></p><pre>curl --location --request POST &#39;https://kafka-rest-cmn.test.tui-dx.com/consumers/sapfi&#39; \</pre><pre>--header &#39;Content-Type:  application/vnd.kafka.v2+json&#39; \</pre><pre>--data-raw &#39;{</pre><pre>&quot;name&quot;: &quot;sapfi_consumer_1&quot;,</pre><pre>&quot;format&quot;: &quot;json&quot;,</pre><pre>&quot;auto.offset.reset&quot;: &quot;earliest&quot;,</pre><pre>&quot;auto.commit.enable&quot;: &quot;true&quot;</pre><pre>}&#39;</pre><p><em>Response</em></p><pre>{</pre><pre>&quot;instance_id&quot;: &quot;sapfi_consumer_1&quot;,</pre><pre>&quot;base_uri&quot;: &quot;http://kafka-rest-cmn.test.tui-dx.com/consumers/sapfi/instances/sapfi_consumer_1&quot;</pre><pre>}</pre><p><strong>Create a subscription for a consumer</strong></p><p>You need to create a subscription indicating the list of topics where you want to consume messages.</p><p><em>Request</em></p><pre>curl --location --request POST &#39;https://kafka-rest-cmn.test.tui-dx.com/consumers/sapfi/instances/sapfi_consumer_1/subscription&#39; \</pre><pre>--header &#39;Content-Type: application/vnd.kafka.v2+json&#39; \</pre><pre>--data-raw &#39;{</pre><pre>&quot;topics&quot;: [</pre><pre>&quot;suppliers&quot;</pre><pre>]</pre><pre>}&#39;</pre><pre>}&#39;</pre><p><em>Response</em></p><pre>HTTP 204 Code: No Content</pre><p><strong>Get the records</strong></p><p><em>Request</em></p><pre>curl --location --request GET &#39;http://kafka-rest-cmn.test.tui-dx.com/consumers/sapfi/instances/sapfi_consumer_1/records&#39; \</pre><pre>--header &#39;Accept: application/vnd.kafka.json.v2+json&#39;</pre><p><em>Response</em></p><pre>[<br>  {<br>    &quot;topic&quot;: &quot;suppliers&quot;,<br>    &quot;key&quot;: &quot;1&quot;,<br>    &quot;value&quot;: {<br>      &quot;id&quot;: &quot;1&quot;,<br>      &quot;customer_code&quot;: &quot;1&quot;,<br>      &quot;telephone&quot;: &quot;888582158&quot;,<br>      &quot;email&quot;: &quot;<a href="/service/mailto:supplier1@test.com">supplier1@test.com</a>&quot;,<br>      &quot;language&quot;: &quot;EN&quot;<br>    },<br>    &quot;partition&quot;: 0,<br>    &quot;offset&quot;: 3<br>  }<br>]</pre><p><strong>Conclusion</strong></p><p>Thanks to Confluent Kafka Rest Proxy, the use of Apache kafka is not exclusive for languages or platforms with kafka clients supported, you can use it through API with easy development and installation as well as with community license.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=555154096540" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/how-to-use-apache-kafka-without-java-client-555154096540">How to use Apache Kafka without Java client?</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Discussing AI, AR, NFTs, Biometrics & Startup Culture (Video)]]></title>
            <link>https://medium.com/tuimm/discussing-ai-ar-nfts-biometrics-startup-culture-video-a86a0b26d918?source=rss----a93729d9156b---4</link>
            <guid isPermaLink="false">https://medium.com/p/a86a0b26d918</guid>
            <category><![CDATA[digital-transformation]]></category>
            <category><![CDATA[traveltech]]></category>
            <category><![CDATA[api]]></category>
            <category><![CDATA[engineering]]></category>
            <category><![CDATA[innovation]]></category>
            <dc:creator><![CDATA[Fabio Zecchini]]></dc:creator>
            <pubDate>Mon, 11 Oct 2021 14:52:49 GMT</pubDate>
            <atom:updated>2021-10-11T14:55:48.345Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*SUWxmX4K9lafDqGnlKWNOg.jpeg" /><figcaption>The leading travel technology event in the world</figcaption></figure><p>I recently had the opportunity to participate in the Travel Tech Show (<a href="https://traveltech-show.com/">https://traveltech-show.com/</a>) on a panel called <em>“The Year When Product Took Centre Stage”</em>, speaking alongside Sundar Narasimhan, SVP of Sabre Labs, and Johnny Thorsen, a serial entrepreneur in the travel tech space.</p><p>We discussed:</p><ul><li>How the pandemic created opportunities for tech teams</li><li>Building startup culture in large organizations (such as TUI)</li><li>Emerging technologies — AI voice interaction, facial recognition and biometric technology, data models, augmented reality, NFTs and more</li><li>Examples of recent tech innovation, creativity and successes</li></ul><p>You can view the full panel discussion video below — feedback welcome.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FI8mNs9vfMBk%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DI8mNs9vfMBk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FI8mNs9vfMBk%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/6aa2292d1cbac6ba201e2d9d91f37d7e/href">https://medium.com/media/6aa2292d1cbac6ba201e2d9d91f37d7e/href</a></iframe><p>As part of the build up to the show, I was interviewed for Travolution.com, discussing how we have changed our ways of working at TUI to become an attractive employer for developers, engineers and those who can play a key role in our digital journey. Check out the interview here: <a href="https://travolution.com/news/in-depth/big-interviews/traveltech-show-how-covid-19-review-laid-the-groundwork-for-tui-musements-future">https://travolution.com/news/in-depth/big-interviews/traveltech-show-how-covid-19-review-laid-the-groundwork-for-tui-musements-future</a></p><p>If you like to travel, and like technology, why not look through the tech focused job vacancies at the world’s leading travel and tourism group: <a href="https://careers.tuigroup.com/digital-technology/">https://careers.tuigroup.com/digital-technology/</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PbfHOD3O5B0pd5feOENdFg.png" /><figcaption>Tech roles at the world’s leading travel and tourism group</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a86a0b26d918" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tuimm/discussing-ai-ar-nfts-biometrics-startup-culture-video-a86a0b26d918">Discussing AI, AR, NFTs, Biometrics &amp; Startup Culture (Video)</a> was originally published in <a href="https://medium.com/tuimm">TUI MM Engineering Center</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>