<?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[Stories by Peter Halacsy on Medium]]></title>
        <description><![CDATA[Stories by Peter Halacsy on Medium]]></description>
        <link>https://medium.com/@halacsy?source=rss-7b3249610585------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*_WgbgjE4wzv24P_e.jpg</url>
            <title>Stories by Peter Halacsy on Medium</title>
            <link>https://medium.com/@halacsy?source=rss-7b3249610585------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 01:29:45 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@halacsy/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Future of JavaScript]]></title>
            <link>https://medium.com/prezi-engineering/the-future-of-javascript-1ad111f5de13?source=rss-7b3249610585------2</link>
            <guid isPermaLink="false">https://medium.com/p/1ad111f5de13</guid>
            <dc:creator><![CDATA[Peter Halacsy]]></dc:creator>
            <pubDate>Thu, 27 Nov 2014 22:33:46 GMT</pubDate>
            <atom:updated>2014-11-27T22:33:46.692Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Note: </strong><em>This post was originally published by TechCrunch on March 15, 2013. It’s reproduced here unaltered, with the exception of the title.</em></p><p>Startups identify with JavaScript. When you’re just starting out, you need to be dynamic. You need to be flexible. You need to be able to bust out a prototype that just works, and you need to be able to change it on a dime without recompiling your code. JavaScript was once the startup of the browser wars, and it crushed Java and Flash for the same reasons that startups have the ability to disrupt markets and displace the established players: agility and flexibility.</p><p>Today, JavaScript is no longer a startup, and it’s no longer just startups that are using it. Companies of all sizes and levels of maturity build applications for the browser, and JavaScript is the language they use to build them. The demands that were once placed on languages like C/C++ and Java are now shifting to JavaScript, and that shift is exposing the language’s limitations: performance and maintainability.</p><p>Those in the latest generation of JavaScript engines deliver incredible performance gains, but these still aren’t enough. Look at sites such as YouTube and Hulu, as well as Facebook’s and EA’s gaming platforms. When it comes to demanding multimedia applications, companies still turn to Flash for performance. The oracle of Steve Jobs may have predicted the ultimate fate of Flash, but we still have to deal with it until JavaScript can match its performance.</p><p>Once your codebase reaches hundreds of thousands of lines of code, and it’s all written in a dynamic language such as JavaScript, development velocity starts to suffer. If you make modifications to one side of your codebase, there’s no way to know if that is going to introduce bugs somewhere else without actually running your application. Testing can only get you so far; there are only so many contingencies that you can anticipate. The result is either more buggy software or longer release cycles.</p><p>Somebody needs to address these limitations so that companies can continue to innovate and build amazing applications for the web. Who’s going to do it and how?</p><p>Prezi, LogMeIn and Ustream organized <a href="http://mloc-js.com/#main">MLOC.js</a>, a conference to focus on how to scale JavaScript. Engineers from Google, Facebook, Mozilla, Groupon and others who are working to solve immense technical problems gathered for three days in Budapest to discuss how they’re tackling the limitations of JavaScript that companies are grappling with today. Their work will ensure that JavaScript is not only a viable platform for the future, but one that will change the way we write applications on the web.</p><p>Here’s my takeaway: JavaScript is caught in an iron triangle of its own making. We’ve all grown accustomed to the flexibility and freedom it offers, and we don’t want to give that up. At the same time, we want it to be blazing fast, and it turns out that when you reach millions of lines of JavaScript, maintaining such a flexible language becomes a problem. All three of these areas — flexibility, performance, and maintainability — are intertwined. When When you make progress in one area, one of the other two, or even both, will suffer.</p><p>This tells me that there is no one silver bullet that will solve these problems, but there may be many that do.</p><blockquote>JavaScript is a dynamic language that excels at being the glue of the web, and that shouldn’t change.</blockquote><p>Business logic needs to work, and it needs to work correctly. The static analysis that systems such as Dart or TypeScript offer is essential to ensuring that the most sensitive components of our applications work — and continue to work as our applications grow and evolve. We need more static analysis tools like these. However, given the browser market, it’s unlikely that we’ll see widespread native support for these technologies any time soon.</p><p>It turns out that native support for new languages isn’t always necessary. Alon Zakai from Mozilla spoke about ASM.js, a subset of JavaScript that runs extremely fast on modern browsers without modification. When speed matters, we should be writing ASM.js. Or even better, we can write our code in C and have a compiler generate the ASM.js for us. Beyond performance, we can apply this technique to other languages and other use cases. Developers are already doing it, and all their efforts are listed on <a href="http://altjs.org/">atl.js</a>.</p><p>One of the best ways to make your code more maintainable is by writing less code. Libraries like <a href="http://baconjs.github.io/">bacon.js</a> or the <a href="http://elm-lang.org/">Elm language</a> succinctly express complex data dependencies in your application and offload the work of keeping your data up-to-date to a library. The result is less code that we have to maintain and higher quality applications.</p><p>JavaScript isn’t going anywhere, but it is undergoing a transformation. Developers are creating language variants and expressive libraries to better handle specific use cases, including performance. They’re making progress on tooling and static analysis, as well. All these trends are changing the way that we write applications for the web. After MLOC.js, that much is certain.</p><p>For the big players out there, the ones who use JavaScript every day, only one question remains: How are you going to contribute to making JavaScript the most flexible, performant, and scalable language we’ve ever known?</p><p><em>Originally published at </em><a href="http://engineering.prezi.com/blog/2013/06/06/the-future-of-javascript/"><em>engineering.prezi.com</em></a><em> on June 6, 2013.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1ad111f5de13" width="1" height="1" alt=""><hr><p><a href="https://medium.com/prezi-engineering/the-future-of-javascript-1ad111f5de13">The Future of JavaScript</a> was originally published in <a href="https://medium.com/prezi-engineering">Prezi Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How and why Prezi turned to JavaScript]]></title>
            <link>https://medium.com/prezi-engineering/how-and-why-prezi-turned-to-javascript-56e0ca57d135?source=rss-7b3249610585------2</link>
            <guid isPermaLink="false">https://medium.com/p/56e0ca57d135</guid>
            <dc:creator><![CDATA[Peter Halacsy]]></dc:creator>
            <pubDate>Tue, 01 Jul 2014 15:24:07 GMT</pubDate>
            <atom:updated>2014-07-06T16:14:32.867Z</atom:updated>
            <content:encoded><![CDATA[<h4>and why it took four years</h4><p>Today we updated the Prezi player, which you can use to view and present your prezis. It is a huge change, completely moving the underlying technology from Flash to Javascript. Hopefully, nobody notices. It has been the biggest migration project of our working-lives, and has taken us almost four years to complete. With its release, I’m pretty sure that Prezi is among the top 1% of the most complex JavaScript applications on the net.</p><p>In celebration of such a time-consuming and difficult project, I have composed an indulgent and long blogpost that attempts to address multiple audiences. Because I am a sympathetic person, I have included large readable headlines for those with limited time or interest. Those with time to kill and a healthy amount of curiosity should scroll down slowly and read the whole thing. #oldschool</p><h3>A topical opening comparison</h3><h4>(which might help also boost our SEO traffic to this post)</h4><p>The World Cup is upon us. At its worst, the competition is a tribal celebration of nationalism centered around a game I don’t care for. At best, its a great excuse to drink with fantastic Prezi people from twenty-four countries in glorious late evening Budapest sunshine (<a href="http://prezi.com/jobs">we’re hiring</a>). During one of these evenings, I heard a sentence from an English colleague that intrigued me:</p><blockquote>“You can tell when the referee has done well if at the end of the game you didn’t notice him.”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n-PhMkbI2ZhvP0Myx7wdNQ.jpeg" /></figure><p>A top-class referee will on average run up to 12 miles during a 90-minute match. That’s five miles more than most players, who are often half the age of a typical official. Imagine working so hard in order to go unnoticed while those around you revel in glory. For a group of Prezi engineers, no such imagination is required. Over the last 5 years, I have essentially been asking some of our top people to do the exact same thing–work really hard in order to go completely unnoticed. This is counter-intuitive for a company that’s all about making your ideas stand out, and having people take notice of you. But for the release of our JS Viewer, our most significant technological advancement since our public release, the key metric for the team is that most people don’t notice a thing. Actually, that’s not quite true, it’d be perfect if some people thought, <em>“Oh Prezi seems a bit smoother than before”.</em></p><p>You can judge the results for yourself, but after having spent the past five years working on this release, I’m now delighted to say that we have a JavaScript player that works from your browser.</p><h3>The part where I thank people and comment on the current state of JavaScript vs Flash</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W3PSDijoeo95MsRhWAPO1g.jpeg" /><figcaption>Flash, Dzso, Adam, Vuji, Loci, Sipi, Gazso, Lacika, Peti, Mate, Zozo, Greg, Zsol, Kalman, Rishabh, Matyi, Gida, Simone, Balint, Attila, Jeremy, Czapo, Andras, Csaba, Omar, Andras! Good job trying something new and different — I know you haven’t done this before.</figcaption></figure><p>I am extremely grateful to the people who worked on this release. Over the last half a decade, a total of fifteen dedicated and very skilled engineers have worked on this JS player at one point or another. But as important as they are, I cannot neglect to mention the <a href="https://twitter.com/suncho/status/476855789513347072">many thousands more who have been waiting</a> for this release. In April of 2010, the number of anticipators grew significantly when <a href="https://www.apple.com/hotnews/thoughts-on-flash/">Steve Jobs declared the death of Flash</a>. Before he did this, HTML5, JavaScript, etc (the names and trends keep changing, and I gave up tracking which one is cool right now) were pretty much just for hipster coders and anyone who liked to think they were better than the mainstream with no interest in producing usable results. But with Steve’s open letter, all of a sudden it became an industry-must to dump Flash. Sadly, Adobe’s technology has outlived the prophet of its demise, but now, it might just be (finally) dying. Am I proud that Prezi is part of this “revolution”? Yes. It’s great if everyone thinks we are a sexy company because we work with the coolest technology, even if most of those people don’t honestly know what the difference is. But I’m actually more concerned with the tangible benefits (read more about those below) this gives the people who use Prezi every day. And I’m amazingly proud of how we delivered these results because it was incredibly hard.</p><h3>The bit where I detail the benefits</h3><p>Flash’s competition has reached a stage where it’s actually going to give us a real difference. First, it’s faster, which for Prezi is extremely important. When you’re zooming around a 2.5D canvas, it’s important that you move slowly from one point to the next.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*RqidQmarM7pQMGMi5TqGGA.gif" /></figure><p>In Minority Report, Tom Cruise didn’t have to wait while the screen jerked around. Instead he swipes and glides around his interface, and people expect the same fluidity from Prezi. The move to JavaScript isn’t going to bring us one step closer to Pre-crime, but this version should make everyone’s presentations smoother. Also, while Tom looks great with those gloves on, I don’t think that giant screen is particularly easy to transport and herein lies the true potential of the technology shift: we believe this is going to lead to a seamless Prezi experience across all devices.</p><h3>The part where I complain about how life isn’t fair.</h3><p>The great thing about Prezi is that it zooms. Zooming is unequivocally awesome. Why? Because it illustrates relationships. Turns out that is important; if you want audience to understand and remember your message, which most people do, showing people your idea AND how it connects with your story at large by zooming from <em>A </em>to <em>B</em>, (or zooming out to reveal both <em>A </em>and <em>B) </em>increases your onstage effectiveness. This is why companies like the Crunch Fitness Franchise claim to have closed 30% more sales since switching to Prezi.</p><p>The difficult thing about Prezi is that it zooms. Zooming makes everything exponentially harder, for engineers. Let’s say I was working for a generic photo sharing platform. I should imagine that one of my main problems is making sure a photo looks as good on a 5-inch screen as it does on a 12-inch one. Prezi engineers have the same problem, and then the user asks to zoom in. In less than a second, we now have to present the same content from a different distance, and all the time make sure the projected image remains rich, sharp, and focused. And we have to do this while multiple people simultaneously view the content from various points around the globe using different browsers and operating systems. So if you’re asking how come <a href="http://engineering.slideshare.net/2011/09/slideshare-ditches-flash-for-html5/">SlideShare managed to release an HTML5 version in 2011</a> and we’re only doing it now in 2014, this is my answer: <em>We zoom, they don’t.</em></p><h3>More detail — WARNING: Probably only interesting to geeks.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/627/1*g7wbTI1HhsAvZzfDJY8QIQ.png" /><figcaption>The problem we had to solve: different browsers render texts slightly differently</figcaption></figure><p>Let’s take the example of font rendering. Did you know that different browsers use different font rendering algorithms? At first glance, nobody recognizes the minor differences between one and the other. When you zoom in on the text those differences get amplified. As an extra challenge, we also have a native mobile version that must be pixel-perfectly identical to the JS version. So we ended up creating our very own text renderer. Just like Flash did 15 years ago.</p><h4>Visual productivity tools are hard</h4><p>Animation performance is crucial for anyone using prezi. Nobody wants jerky prezis. So we had two engineers fine-tune the rendering of large amounts of content to the canvas in the last four months. Yes, we are using Canvas. Bitmap caching of small but complicated drawings had the biggest impact on the median frame rate. Thanks to Flash, we already had some experience in optimizing the speed of rendering on a canvas-like platform.</p><blockquote>“The key learning for us has been that dedication, hard work, and optimization of how you are using Canvas can yield hard performance animation even in HTML5.”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/731/1*n4m9sb-JfP1ic25fxTh3Lw.png" /><figcaption>One of the charts we use to monitor the speed of rendering.</figcaption></figure><p>The first steps are always: set up and automate the right measurements, and then monitor the speed.</p><p>Prezi also needs high performance drawing of <strong>vector graphics. </strong>Nobody wants a pixelated chart or logo, nobody. Flash was really good at vector graphics, JS is not as good (we know about SVG; if you want to know why SVG was not the best solution for us, feel free to <a href="http://hu.linkedin.com/in/peterhalacsy">invite someone</a> from the team to come and talk at your conference or meetup.). So we ended up creating our own vector graphics renderer based on <a href="http://mozilla.github.io/shumway/">Shumway</a> (yes, we are going to publish all of our improvements, Mozilla will merge soon).</p><p>The biggest challenge was <em>adding clipping to Shumway. </em>I guess nobody thought about how Shumway could be used to render small parts of huge vector graphic files (like PDFs). But our users do like to zoom in…</p><h4>Large JS codebases are hard</h4><p>Besides zooming there was one other big challenge: We have a pretty enormous codebase compared to the average web-based JS application. If you don’t agree that a large codebase can be challenging, then read <a href="http://techcrunch.com/2013/03/15/the-future-of-javascript/">this TechCrunch article</a>.</p><p>When working on a large code base, we prefer typesystems and compilers. That’s why we write the majority of our code in <a href="http://haxe.org/">Haxe</a>. We are big Haxe fans. For your curiosity: we are also experimenting with Typescript. And yes, we are using <a href="https://github.com/kripken/emscripten">Emscripten</a> to compile some code of the text renderer from C. Regardless, we are a <a href="http://engineering.prezi.com/blog/2014/01/20/prezi-the-land-of-the-polyglots/">language agnostic group of people</a>, so we support any X-to-JS language.</p><blockquote>“We invested a huge amount of engineering time in building our own toolset, something I really recommend to all teams working on large applications.”</blockquote><p>We are big fan of microcomponents (we built the app from 18 independent components), automatization, and devops like deployment.</p><p>We created and open-sourced our own <a href="https://github.com/prezi/spaghetti">module system</a> and <a href="https://github.com/prezi/pride">build system</a> based on Gradle. Auto-scaled Jenkins nodes are running the tests on 7 browsers after every commit. We automatically publish all versions to the production system and make them available for everyone. Of course, fine grained feature switches make selective rolling out possible. The whole tooling system is build to increase developer happiness. And could make you happier than improving the tool used by millions. That’s why it’s important to lower the time it takes from changing the code on a local machine to making it available for people to use. At Prezi, this takes only 15 minutes.</p><h3>So why only in 2014?</h3><p>We launched Prezi in 2009? iPhone without Flash appeared in 2010. Why didn’t we release a JS version earlier?</p><p>It was not our key focus until 2013. We have had a native iOS version since 2011 and our users seems to be happy with Flash. Still sounds like bullshit? Here’s the other reason:</p><p>We tried two really bad approaches before this project.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/841/1*VwChBopM9Alxi5w6Wmy6Sw.png" /><figcaption>The vanialla HTML5 version using DOM + CSS3. It never went to production.</figcaption></figure><p>First we tried to use vanilla HTML5, <a href="http://intra.prezi.com/~adam2/prezi/index.html">DOM + CSS3 animations</a>. Cool, huh? Nope. Not cool. We never could turn this version into a high-quality enough product. So we shifted the focus to a native iOS version.</p><p>Our <a href="http://www.silexlabs.org/135169/the-blog/multiplatform-productivity-tool-with-haxe/">second project aimed at having a full multi-platform codebase</a>: generating JavaScript and native apps from the same codebase. Why? Because we always wanted to have 100% pixel-perfect identical prezis on every platform. It took us almost a year to realize that the goal is important but our solution was a really bad idea.</p><p>Today the JavaScript and native teams are totally separated. Both use their own toolsets, processes and environments. The key learning?</p><blockquote>“In the long-term, using a popular platform’s standard processes and toolset results in a higher quality product and happier engineers. Don’t fantasize about using an obscure toolchain to <a href="http://en.wikipedia.org/wiki/Write_once,_run_anywhere">write-once</a> a small core codebase. Engineers actually love to write code. “</blockquote><h3>For managers</h3><p>If you’re this far in, you’re either a dedicated reader, or someone who scans through the headings for the parts you’re interested in. Either way, I’ll get to the point.</p><p>In all honesty, we have been working on this for a long time. As I alluded to, we flirted with the idea since 2010. During this time, we overcame the various difficulties by being what I call “truly agile”. I like to use this combination of words because it makes me feel like some kind of influential thought-leader. In reality, it means some quite simple adjustments to the way you work. These are (in order of importance):</p><ol><li>Generate enough money to hire people.</li><li>Hire the right people.</li><li>Give them the right infrastructure and culture to succeed.</li><li>Trust them.</li><li>Dedicate them to one task (in this case the technology shift).</li><li>Give them clear accountability.</li><li>Always have an “almost ready” version. This will force them to think about what they have to do instead of what they’ve already done.</li></ol><p>There’s also something to be said for ensuring you learn from mistakes, but this is part of hiring the right people dedicated to the right task working under the right culture.</p><h3>For journalists</h3><p>I could deny that we’re working to create a JavaScript/HTML5 editor for Prezi, but I won’t. You only need take a look at our jobs page (did I mention <a href="http://prezi.com/jobs">we’re hiring</a>), and you can see this is what we’re working on. If you’ve read the rest of this piece, you’ll know how hard it was to get this far (if you didn’t, scroll up), and as such I won’t be promising when this will happen, just that I’m confident that we are on the right track to getting there.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=56e0ca57d135" width="1" height="1" alt=""><hr><p><a href="https://medium.com/prezi-engineering/how-and-why-prezi-turned-to-javascript-56e0ca57d135">How and why Prezi turned to JavaScript</a> was originally published in <a href="https://medium.com/prezi-engineering">Prezi Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Prezi — the Land of the Polyglots]]></title>
            <link>https://medium.com/prezi-engineering/prezi-the-land-of-the-polyglots-c87144b23093?source=rss-7b3249610585------2</link>
            <guid isPermaLink="false">https://medium.com/p/c87144b23093</guid>
            <dc:creator><![CDATA[Peter Halacsy]]></dc:creator>
            <pubDate>Thu, 26 Jun 2014 08:38:08 GMT</pubDate>
            <atom:updated>2014-06-26T08:38:08.148Z</atom:updated>
            <content:encoded><![CDATA[<p>“I speak Spanish to God, French to women, English to men, and Japanese to my horse.” —Buckaroo Banzai</p><p>Ask any engineer “which programming language is the best?”, and you’ll get crushed by a deluge of answers ranging from stuff they started with in uni to something they read about on a tech blog ten minutes ago. Some will answer with whatever they’re using right now in their job, and a few more will tell you about something they think is going to be big, if only Google/Microsoft/Apple/King would see it for what it is!</p><p>We’re living in a multi-platform age. Gone are the days of writing for one operating system, in one or two languages, for only one (primary) client. Take Prezi. We never hire anyone as an “expert” Flash (or even JS) coder, but our product runs primarily on Flash, so a lot of people become one. Nobody really gets to specialize — one week you’re working in C++ (<a href="http://prezi.com/jobs">and we’re looking for you if you can too</a>), and next week maybe the team needs you to dig down into JavaScript to get the job done.</p><p>Being an engineer/dev into today’s world means being a polyglot. At Prezi, we’ve got at least 14 main languages under our collective belts that people actively work in, day in and day out.</p><h3>What We Do</h3><p>We have a diverse client base running on several platforms. We’re gunning for an “as-native-as-it-can-be” user experience, so we want to use the best tools we can to achieve that. We want you to be able to say “Hey, Prezi’s pretty cool and works on my machine!”. That gets us our first five languages:</p><ul><li>On our browser and desktop clients, our main language is <strong>ActionScript</strong>, as this is the language of Flash. This serves 95% of our active users. So, we gotta do it.</li><li>On the iOS we use <strong>ObjectiveC </strong>as this is the language of iOS. So we use it.</li><li>On Windows, <strong>C# </strong>or <strong>C++ </strong>are the most supported languages. Since that’s the desktop OS for a whole lot of our users, we use it too.</li><li>Now that we’re delving into Android, we need <strong>Java </strong>too, obviously.</li><li>For modern browsers, we use <strong>JavaScript. </strong>The core of the Prezi experience is (at least for now) a cloud-based presentation service. Web applications use JavaScript (and CSS and HTML of course) these days. Unless you live 25 years in the past. But then you’re trying to figure out which hand isn’t too tired to hold your brick-sized phone, or where you’ll get a pencil to rewind your cassette tape by hand.</li></ul><p>Because we’re a primarily cloud based service platform, we need some server side languages too (that’s six and seven):</p><ul><li>Historically our server code is in <strong>Python </strong>which is one of the most productive general languages.</li><li>For some sophisticated server code we use <strong>Scala (</strong>JVM). This lets us reuse industry standard libraries (especially <em>Netflix </em>open source stack) and Java is one of the most well-known, reliable technologies. It works, and everybody knows it.</li></ul><p>We also use <strong>domain specific languages </strong>for some specific use cases. The Prezi cultural spirit, for each team to be as cross-functional as possible, suggests that almost all engineers should learn to use these technologies (and now eight, nine, and ten):</p><ul><li>We use multiple DSLs written in <strong>Ruby</strong>; <em>Chef </em>for configuration management, and <em>Cucumber </em>for acceptance tests.</li><li>We use <strong>Pig Latin </strong>for ad hoc data mining (<em>Hadoop) </em>and some level of <strong>SQL</strong>for general data retrieval</li><li><strong>Shell scripting </strong>knowledge is common sense in Prezi because using the command line is often more effective for certain tasks, and you have to get comfortable with it.</li></ul><p>This lets us actually serve our client base in a way that might give them something they need and want.</p><h3>Drive for Efficiency</h3><p>Besides those first six we are experimenting with less mainstream languages which could drastically increase our efficiency. It’s a common approach in startups to choose a more efficient tool for their tasks (as opposed to, say, picking ONE language to use, for everything, forever, even if nobody likes it, because one particular boss said so).</p><p>For example, fellow startup <em>Dropbox </em>invested in Python when it was an unknown and rarely used language; <em>Twitter </em>migrated everything to Scala (basically triggering Scala’s current popularity).</p><p>What we’re playing (and working) with (eleven, twelve, thirteen, and fourteen baby!):</p><ul><li>We are using <a href="http://haxe.org/"><strong>Haxe</strong></a><strong> (</strong>an open source high-level multi-platform language and compiler inspired by ActionScript and C#) to write some of our JavaScript and Actionscript code, because it gives us a stronger type system to help us find bugs during compile time. Generally speaking, we believe in the Generate to Javascript approach when we use JavaScript as an assembler. This approach is followed by some Google projects (GWT, Dart), Microsoft (TypeScript, C#2JS) and even Firefox invested in the opportunity (Asm.js, emscripten). For more on this read my contraversial post <a href="http://techcrunch.com/2013/03/15/the-future-of-javascript/">“Is The Death Of JavaScript Upon Us”</a> at TechCrunch.</li><li>On the server side, we are moving from Python to Scala (and <strong>Kotlin) </strong>to increase our efficiency. Scala code is shorter and easier to maintain.</li><li>We are experimenting with code generation and domain specific languages. Why? Because we want to be sure that if we do write a component in 5 languages, it works the same everywhere, consistently; code generation (for a DSL) improves efficiency. For one of our projects, we use <strong>Haskell </strong>that makes parsing and symbolic manipulation easier.</li><li>We are the sole sponsor of an open source language called <a href="http://elm-lang.org/">Elm</a> . The promise of Elm is that it will make development of highly interactive web applications more efficient (and since we’re a web application…).</li></ul><h3>14 languages. How on earth?</h3><p>So, does everybody have to learn everything? No, of course not. That’d be silly. We’d spend all our time learning coding languages and while I might like paying people to learn, I’d rather they learn by actually working on our product. And we’ve figured out what every other company already knows: it’s nice to specialize a little.</p><p>Do we waste time on supporting all languages? Again, no. But we do support our engineers taking time to learn a new one. Learning a new language isn’t that hard once you get past your first one or two. Learning a new paradigm (like OO or FP) will develop your brain and will make you a better engineer in general. Modern tech companies have to forget the idea that learning a new programming language is like learning a new communicative language (something horrible, slow, and expensive and time consuming). This is more like learning a new dance. You already know how to Tango and you’ve never done Irish Stepdancing or Maori Haka, but they still use all those muscles and skills you trained up before, just in new ways.</p><p>Our model engineer is in more of a polygon. If you work on a server side project, for example, then Scala/Java/Shell scripting with some Python/Ruby is enough for your different “sides”. Some of the polygons are smaller than that. Maybe one of our engineers only needs Javascript/C++ with some Objective C to give them the result they want. Or not. It depends.</p><p>Which is my last point really. We want our people to use the right tool for the right job. If a team figures out that using Perl is the way to go for a particular project, then, as long as they take responsibility for it (after all, maybe other teams can’t help if they’re not conversant in Perl) then, why not? So it depends on what you’re doing. What we don’t want, and will fight against until our last breath, is getting stuck in the one-way rut. We want our people learning new things while they’re doing the old things they always do. We want minds stretched and assumptions on how to do things challenged. In other words, we want Prezi engineers to be able to grow. And that’s something that’s going to be the theme for all tech companies as we move into 2014.</p><p>(Honorable mention: there are a LOT of other languages that get used, or that certain people or teams like, or that we find fun and instructive to try out, stuff like: Groovy, Kotlin, Clojure, and R. That’s what this article is about; trying out stuff to see what improves us as a business, and as individuals. There’s just too many to list…)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c87144b23093" width="1" height="1" alt=""><hr><p><a href="https://medium.com/prezi-engineering/prezi-the-land-of-the-polyglots-c87144b23093">Prezi — the Land of the Polyglots</a> was originally published in <a href="https://medium.com/prezi-engineering">Prezi Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>