nik's site - Blog posts

Adding autotracking to a security camera

2025-11-07T00:00:00+00:00

I recently got my hands on a generic Chinese camera.</p>

The App Situation</a></h2>
This camera requires an extremely shady mobile app (V380 Pro — Google Play</a>) to function. This is easily among the worst pieces of software I've ever interacted with. Slow, janky, and filled to the brim with ads — both regular ads, and promotion for their "cloud service" to store (among other things) camera footage.</p>
I spent some more time in the app, looking at the capabilities that it exposes.</p>

It "implements" PTZ (Pan, Tilt, Zoom), but zoom doesn't work.</li>
It can save/load preset</em> positions.</li>
It has a mode where it's supposed to autotrack</em> people/objects. Funnily enough, the camera rotated in the opposite</strong> direction of the object when I tested this.</li> </ul> A man, a plan, a lan</a></h2>
I wanted to do the following things:</p>

Make the camera usable without the shitware app.</li>
Try to fix/implement autotracking.</li>
Prevent it from leaking my data all over the place.</li> </ol>
When I dug into the settings in the app, I discovered a redeeming quality: this wonderful little wiretap supports ONVIF (for future me, ONVIF port was 8899).</p>
ONVIF is a standard for controlling networked cameras. This is great news! I can implement the person-tracking functionality on my own!</p>
I messed around with Frigate</a>, trying to get its autotracking to work with the camera. However it just refused to work. Pan/Tilt worked perfectly, but autotracking did not.</p>
I decided to take the matter into my own hands and see if I could implement autotracking on my own.</p>
I made the following plan:</p>

Write two scripts:

One that exposes a function that can be used to move/reset the camera.</li>
Another that fetches the video stream from the camera, uses OpenCV to do the detection, and move the camera so that it points straight at the target. Maybe record events?</li> </ul> </li>
Drop all this onto a Raspberry Pi.</li>
Use the Pi's hotspot to connect to the camera, with very restricted internet access.</li>
Set up a Wireguard VPN to connect to this little network from outside.</li> </ul>
This post will cover just the first part of this glorified plan.</p>
Autotracking kinda works!</a></h2>
Tracking objects is pretty easy with OpenCV. Turns out, getting the camera to move properly is the hard part.</p>
This is because this camera only implements a small subset of the ONVIF standard, and that too in an annoying way.</p>
(I got to know about this thanks to this script</a> taken from Frigate's docs</a>.)</p>
The camera supports ONVIF PTZ with the following caveats:</p>

It supports only</em> `ContinuousMove</code> (which is basically relative motion — you give it X and Y velocities).</li>`
`The move is executed after</em> you call Stop</code> on it.</li>`
There's no way to query or set the absolute position of the camera.</li> </ol> I (with a lot of help from Gemini 2.5 Pro) wrote the first script, onvif_control.py</code>. It exposes a move(x, y)</code> method that moves the camera in steps</em>. This lets me calculate how much I'm moving with some accuracy.</p> I also wrote the person_tracker.py</code> script that loads the RTSP video stream into a separate thread, and uses OpenCV (with YOLOv3) to recognise people and incrementally move towards them.</p> So far, this is working fine.</p> The biggest problem is that there's no way to return to a starting position. The camera doesn't seeem to expose any way to query the absolute position.</p> I tried storing the steps the camera has taken so that I can backtrack using it, but that doesn't work because move()</code> might not be successful all the time (weird lags with the camera, reaching the end of motion range, etc). It kept throwing my calculation off, so I abandoned it for the time being.</p> I also tried an approach where I essentially "calibrated" the camera by moving it to the extremes of its range of motion and measuring the steps it took to do that, but that didn't work either. This approach still holds some promise though, so I'll probably try this again.</p> My final idea is this: I can maybe reverse engineer the calls the V380 Pro app makes to the camera, and use that proprietary protocol to control it. That's a lot more work, but it seems like it will be more reliable. The app has a way to store PTZ presets, as I mentioned at the beginning. I can probably use that to my advantage if I figure this out.</p> I'll try to clean up and publish the scripts so far if I get time.</p> Stay tuned for a part 2!</p>

keyd: Layers, macros, and remaps, even on laptop keyboards 2025-10-23T00:00:00+00:00 I have a mechanical keyboard (blue switches) with a volume knob. These things have pretty satisfying tactile feedback which makes them fun to play with.</p> I felt that I wasn't using it enough, so I decided that once I get some free time, I'll find a way to remap it to something more useful than adjusting volume.</p> On one lazy day (ie, just before one of the uni exams), I decided to look more closely at the knob. Turns out, when it's rotated clockwise, it simulates pressing the Volume Up</kbd> key, and vice versa counter-clockwise.</p> Now all I had to do was remap these keys to something more useful.</p> Since my keyboard doesn't support QMK</a> or VIA</a>, I had to find something that works on all keyboards.</p> That's when I stumbled on keyd</a>. It's a wonderful little daemon that intercepts keypresses and lets you do things like remap buttons (CapsLk</kbd> to Ctrl</kbd> when held, Esc</kbd> when tapped being a common one), create layers, and more.</p> It works by spinning up a daemon that reads from a config file, after which it happily does its thing in the background.</p> I mapped my volume knob to up/down arrow keys by default, and left/right arrow keys when shift is held.</p> I also created a global config that lets me use AltGr</kbd>+h/j/k/l</kbd> instead of arrow keys (on all keyboards, including my laptop's). I also use AltGr</kbd> as my compose key</a>, which I didn't want to break. Turns out, you can specify that you want to overload</em> a key in the keyd config file. This makes it behave like normal AltGr</kbd> when tapped, but changes to a special nav</code> layer when held. This layer then lets me use my h/j/k/l</kbd> keys for navigating around.</p> This is what my global config looks like:</p> [ids] * [main] # AltGr will be RightAlt/Compose on press, nav layer on hold. altgr = overload(nav, rightalt) [nav] h = left j = down k = up l = right </code></pre> And this is what the config for my mechanical keyboard looks like:</p> [ids] 1a2c:7fff:3c9dd5ab [main] volumeup = down volumedown = up [shift] volumeup = right volumedown = left [control] volumeup = volumeup volumedown = volumedown [alt] volumeup = brightnessup volumedown = brightnessdown </code></pre> Dynamic MathJax typesetting in Svelte 2024-07-31T00:00:00+00:00 (TL;DR;give-me-the-code: Dynamic MathJax rendering — Svelte REPL</a>)</em></p> I am building a quiz platform for the startup I am working on right now. The questions in the quiz often contain LaTeX math elements which need to be typeset by MathJax.</p> The setup</a></h2> MathJax and its config is brought in with the <svelte:head></code> tag, like this:</p> <svelte:head> <script> window.MathJax = { tex: { inlineMath: [ ["$", "$"], ["\$", "\$"], ], displayMath: [["$$", "$$"]], }, svg: { fontCache: "global", }, startup: { typeset: false, }, }; </script> <script id="MathJax-script" defer src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-chtml.js" ></script> </svelte:head> </code></pre> I have an array of questions (which are just HTML text with LaTeX expressions) and a variable called current</code>, which keeps track of the index of the current question. There are also couple of buttons to switch between the questions.</p> <script> const current = 0; const questions = [ `This is $ e^x $: <br /> $$ {\\bf{e}}^x = \\sum\\limits_{n = 0}^\\infty {\\frac{{{x^n}}}{{n!}}} $$`, `And this is $ \\cos x $: <br /> $$ \\cos x = \\sum\\limits_{n = 0}^\\infty {\\frac{{{{\\left( { - 1} \\right)}^n}{x^{2n}}}}{{\\left( {2n} \\right)!}}} $$`, `Finally, this is $ \\sin x $: <br /> $$ \\sin x = \\sum\\limits_{n = 0}^\\infty {\\frac{{{{\\left( { - 1} \\right)}^n}{x^{2n + 1}}}}{{\\left( {2n + 1} \\right)!}}} $$`, ]; </script> <svelte:head> ... </svelte:head> <main> <button on:click={() => current--}>Prev</button> <button on:click={() => current++}>Next</button> <p> {@html questions[current] } </p> </main> </code></pre> </p> The problem</a></h2> MathJax runs the typesetter just once, as soon as it's loaded on the page. At this stage, the question text has not loaded in yet, so the typesetting doesn't take effect.</p> We need a way to make this dynamic</em>, so that MathJax will re-typeset the text when the question number (i.e., value of current</code>) changes.</p> The solution</a></h2> We can use two nifty tools that Svelte provides — the use</code> directive and the {#key}</code> block — to solve this problem.</p> Implementing use:typesetMath</code></a></h3> The use</code> directive</a> calls the attached function (called an action</em>) when the element is created.</p> We can use this to create an action that will check if MathJax has been loaded and run the formatter (window.MathJax.typesetPromise</code>) on our element.</p> <script> // ... // // Action functions accept the HTML element as an argument when they are called function typesetMath(node) { const typesetPromise = window.MathJax.typesetPromise; // Check if MathJax has actually loaded before calling `typesetPromise` if (typesetPromise) typesetPromise([node]); } //...// </script> ... <p use:typesetMath>{@html questions[current] }</p> ... </code></pre> Now, we need to make sure that this action is fired whenever the variable current</code> changes.</p> Setting up the {#key}</code> block</a></h3> The {#key}</code> block will cause the elements inside it to re-render whenever the key</em> (in this case, the variable current</code>) changes, which then goes ahead and triggers the typesetMath</code> action.</p> {#key current} <p use:typesetMath>{@html questions[current] }</p> {/key} </code></pre> So far so good! Everything should work more-or-less as expected now, but there is one little bug.</p> The typesetting might not happen properly for the question that is shown by default, because MathJax might not have been loaded by the time the typesetMath</code> call is made, and the function exits without doing anything.</p> </p> To fix this, we can use the else</code> branch in the conditional in typesetMath</code> to call itself recursively (with a small delay) until MathJax is loaded.</p> function typesetMath(node) { const typesetPromise = window.MathJax.typesetPromise; if (typesetPromise) typesetPromise([node]); // recursively call itself every 500ms until MathJax loads else setTimeout(() => typesetMath(node), 500); } </code></pre> And that's it! That's how I hacked it together. It is not an elegant fix but it works pretty well for now.</p> </p> Click here for the full code. </summary> <script> import { afterUpdate } from 'svelte'; let current = 0; const questions = [ `This is $ e^x $: <br /> $$ {\\bf{e}}^x = \\sum\\limits_{n = 0}^\\infty {\\frac{{{x^n}}}{{n!}}} $$`, `And this is $ \\cos x $: <br /> $$ \\cos x = \\sum\\limits_{n = 0}^\\infty {\\frac{{{{\\left( { - 1} \\right)}^n}{x^{2n}}}}{{\\left( {2n} \\right)!}}} $$`, `Finally, this is $ \\sin x $: <br /> $$ \\sin x = \\sum\\limits_{n = 0}^\\infty {\\frac{{{{\\left( { - 1} \\right)}^n}{x^{2n + 1}}}}{{\\left( {2n + 1} \\right)!}}} $$` ] function typesetMath(node) { const typesetPromise = window.MathJax.typesetPromise; if (typesetPromise) typesetPromise([ node ]); else setTimeout(() => typesetMath(node), 500) } </script> <svelte:head> <script> window.MathJax = { tex: { inlineMath: [ ['$', '$'], ['\$', '\$'] ], displayMath: [['$$', '$$']] }, svg: { fontCache: 'global' }, startup: { typeset: false } }; </script> <script id="MathJax-script" defer src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-chtml.js" ></script> </svelte:head> <main> <button disabled={current===0} on:click={() => current--}>Prev</button> <button disabled={current===questions.length-1} on:click={() => current++}>Next</button> {#key current} <p use:typesetMath> {@html questions[current] } </p> {/key} </main> </code></pre> </details> I have also created a Svelte REPL</a> if you want to play around with this.</p> Escaping the Doomscroll 2024-06-24T00:00:00+00:00 (Warning: Ramble ahead!)</p> You probably know at least one person who spends hours and hours scrolling on their phone, most probably consuming content that provides barely anything of entertainment value, let alone something actually stimulating. Maybe you</em> are one of those poor souls who have sold their souls to The Algorithm, succumbing to its infinite feed.</p> Algorithmically curated, infinitely-scrolling feeds are an unfortunate staple of pretty much all the major “social” media platforms (henceforth called “antisocial media” because that’s more accurate in my opinion). As you probably already know, these antisocial media platforms make money by showing ads. Targeted ads. And how are they targeted? By carefully analysing every little move you make and every word you say on their platforms and beyond, and building a profile out of it that knows more about you than your closest family does. Data mining is scary</a> but that’s a discussion for another day.</p> It is obvious that it is in the antisocial media platforms’ best interests to keep you on their platforms for as long as humanly possible, and what better way to do it than target the novelty-seeking nature of our brains!</p> Users can’t really be blamed for being addicted to this infinite-scrolling conveyor belt to hell, as these are carefully crafted by psychology and neuroscience experts to keep you hooked for as long as possible. Addiction is not a bug here, it is the biggest feature.</p> Facebook, Twitter, Instagram, etc</a></h2> I am happy to say that I never had an account on Facebook or Instagram, and I won’t be making any in the near future if I can help it. Even if I do make an account due to real-life social reasons (unfortunately that is a possibility thanks to the network effect, the same reason why I use WhatsApp) I believe that I know better than to scroll on it all day and progressively ruin my general mood, as I’ve seen so many people do.</p> I did make a Twitter account once, just out of curiosity. I don’t think I have access to it or the email I used to create it any more. I barely used it for a day so I don’t think it really counts either.</p> Seriously, the best option is to simply never touch these unless you absolutely</em> have to. (eg: promoting a business, unreasonable acquaintances)</p> TikTok</a></h2> Brainrot. Banned in my country.</p> Although the moral implications of the government exercising control and banning software from app stores are pretty chilling, I’m happy to say that TikTok will not be missed. At all.</p> YouTube Shorts</a></h2> Several apps tried to fill the pathetic void left by TikTok in my country. The only one that actually succeeded was YouTube with its YouTube Shorts. This steaming pile of garbage is still pushed very aggresively in the official YouTube app.</p> I had switched to NewPipe</a> long before YouTube Shorts was even a thing. Shorts show up like normal videos on the NewPipe app, as opposed to the infinite strand of spaghetti in the official YouTube app that brings you closer and closer to the Singularity with every scroll.</p> Miscellaneous</a></h2> I had an account on Reddit, and I had a tendency to doomscroll on it. However the Reddit API changes that broke third-party clients cured my addiction. The official Reddit client is so broken and unusable that it’s impossible to get addicted. Good job, Reddit!</p> I do have an account on Lemmy and Mastodon. The Lemmy one was made right after the Reddit API disaster and I use it to occassionally look at some memes. Lemmy is still pretty small, so it doesn’t take me long to exhaust the supply of memes and I can put my phone down afterwards. I barely use Mastodon. It’s too messy for my taste.</p> Conclusion</a></h2> If you have the attention span to read all the way until here, you’re doing just fine. Your brain hasn’t completely decomposed yet and your soul is still within you and has not been sucked away into the depths of some algorithmic feed.</p> Thanks for reading!</p> Adding a search engine to Firefox 2024-05-23T00:00:00+00:00 Search engines in Firefox are specified by either OpenSearch</a> tags or Firefox Add-ons. This makes it really difficult to add site-specific search engines that don't support either of these (like Hacker News's site search</a>).</p> This is quite different from most Chromium-based browsers, where you can add any site-search engine by specifying its URL with a %s</code>, which is then substituted with the actual search query.</p> However there is a way to replicate this behaviour in Firefox, although it is hidden behind a flag and not exactly straightforward to enable.</p> This is how you can enable it in Firefox:</p> </p> Open about:config</p> </li> Type in the following in the search bar:</p> browser.urlbar.update2.engineAliasRefresh </code></pre> </li> Make sure that boolean</code> is selected, click on the + button, and set it to true</code></p> </li> </ol> Done!</p> Now, when you navigate to Settings > Search (about:preferences#search), you'll find a new “Add” button under the Search Shortcuts list. Just click on it to add a new search engine.</p> </p> The Compose Key: Typing alternate characters the easy way 2024-05-19T00:00:00+00:00 (Note: I've mostly used UK/US English keyboards and my article might be biased towards these layouts.)</em></p> How do you type the capital letter H?</p> I think it is safe to assume that you press Shift</kbd> h</kbd>, instead of navigating to the Wikipedia page for Horse</a> and copying “H” from there.</p> How do you type, say, the ñ in España, or the ç in Françoise?</p> You don't have to visit their respective Wikipedia pages and copy the word from there like I used to do.</p> Just use the compose key.</p> Compose key? I can't find that on my keyboard!</a></h2> You might have to download some additional software to set up compose key on Windows</a> or macOS</a>.</p> On most Linux desktops, however, it is pretty straightforward. You might find these guides useful: KDE Plasma</a>, GNOME</a>, XFCE</a>.</p> (For reference, I'm on KDE Plasma and I have my AltGr</kbd> mapped to Compose</kbd>.)</p> With the compose key, ñ becomes Compose</kbd> n</kbd> ~</kbd>, and ç becomes Compose</kbd> c</kbd> ,</kbd>.</p> Compose key mappings are pretty intuitive. Let's say I want to type this: ø. Now what does that look like? An 'o' with a slash through it? Yep! Compose</kbd> o</kbd> /</kbd> will give you a ø.</p> Compose key mappings</a></h2> You can create your own compose key mappings by defining them in the ~/.Xcompose</code> file, but I have personally found this to be a bit finicky.</p> Some default mappings that I often use</a></h2> Superscript numbers: Compose</kbd> ^</kbd> 8</kbd> for ⁸</li> Smart quotes</a>: Compose</kbd> <</kbd> "</kbd> for “</li> Compose</kbd> ></kbd> "</kbd> for ”</li> Compose</kbd> <</kbd> '</kbd> for ‘</li> Compose</kbd> ></kbd> '</kbd> for ’</li> </ul> </li> Dashes Compose</kbd> -</kbd> -</kbd> -</kbd> for — (an em dash)</li> Compose</kbd> -</kbd> -</kbd> .</kbd> for – (an en dash)</li> </ul> </li> Compose</kbd> <</kbd> 3</kbd> for ♥</li> Compose</kbd> :</kbd> )</kbd> for ☺</li> Compose</kbd> :</kbd> (</kbd> for ☹</li> </ul> Further reading</a></h2> https://en.wikipedia.org/wiki/Compose_key</a></li> https://github.com/kragen/xcompose</a></li> http://canonical.org/~kragen/setting-up-keyboard.html</a></li> </ul> On writing 2024-04-17T00:00:00+00:00 Categorising writing</a></h2> I noticed that my writings fall on a gradient, whose two extremes are:</p> Writing purely for myself</li> Writing purely for someone else</li> </ol> On one end of the spectrum, I have things like journals and notes and todo lists and the like. On the other end, I have academic pieces and tutorials.</p> This blog falls squarely in the middle.</p> Writing as an emotional outlet</a></h2> Journalling. It really makes me calm down and allows me to reflect on things properly.</p> Things almost always seem bigger than they are when you are dealing with them in your head in real time.</p> I discovered that if something is troubling me, I can just write it down and come back to it a few hours later. When I look at it again after a few hours, the problem will have magically become smaller and more solvable.</p> It keeps my sanity in check (to an extent, of course).</p> I don't use anything fancy for journalling either. Just Markdown files.</p> I did consider a paper journal but it's not easy to back those up, plus typing is faster anyway.</p> Writing a blog</a></h2> Actually writing a proper blog is a new experience for me. Lately most of my work has been on the extremes of the categorising that I described at the beginning (Writing purely for myself or writing purely for someone else).</p> In the first case, I have complete freedom of expression without regards to any consequences, and I can write without having to think about style or consistency or whether I will offend someone. It is an open game and I am the only player. But because of obvious reasons, I cannot share or publish it.</p> In the second case, my writing invariably becomes drab and depersonalised and as painful to read as it is to write.</p> I wanted to see if I can strike a balance between the two. I want to improve the way I present my thought processes to others. I feel like this is a very essential skill, and I have a lot of room for improvement in this department.</p> I feel that one's ideas, no matter how great they might be, are ultimately useless if no one else can understand it.</p> How I write</a></h2> I use one of those cheap LCD tablets</a> for more short-term content, like diagrams for brainstorming. (And also for solving Physics problems :-P)</p> For anything more concrete, I use my computer.</p> I typically write in Markdown using Neovim. As I can type faster than I can write with a pen on paper, it happens to be by far the fastest and most efficient way I've found to get large amounts of text out of my brain and onto a more permanent storage medium.</p> Sometimes, when I'm working on something, I get ideas for topics that are not quite relevant to what I was working on. I usually throw these to a file called something like ~/ideas.md</code>, and review them when I get time.</p> I've found this to be an excellent strategy to not lose good ideas, while also being able to focus on what I am doing instead of chasing the shiny new idea.</p> Speaking of which, I should also write a post on distraction. More specifically my concept of “return-on-distraction”.</p> $ echo "Write blog post about return-on-distraction" > ~/ideas.md </code></pre> Added to the list!</p> Setting up Zola for fun and profit. Also, hello world! 2024-04-17T00:00:00+00:00 I've made a lot of blogs, but I never really maintained any of them as I would lose interest after the first couple of posts or so.</p> Let's hope things are different this time!</p> With that out of the way, let's talk about the cool stuff.</p> Zola</a></h2> This blog (as of writing) is built on the Jamstack</a>.</p> I am using a static-site generator called Zola</a>. I write my posts in Markdown, and Zola converts them to HTML pages using some templates. And it's fast</em>. Right now, with only one post, it takes just 6ms to generate the whole site!</p> I've used Hugo</a> and 11ty</a> in the past, but they felt a bit overkill for my little blog and I wanted to try something different and preferably a bit more minimal.</p> I was actually planning to cobble together my own static-site generator using pandoc</code> and some shell scripts, but VicFic</a> saved my sanity and suggested Zola.</p> So, I followed Zola's getting-started guide and set things up. I wrote my own CSS and I ported over my old homepage to Zola. It was pretty smooth and less things went wrong than I had anticipated.</p> Styling</a></h2> I'm using SCSS</a> to write the stylesheets. The font is IBM Plex Mono</a> which I got from xz/fonts</a>.</p> Hosting</a></h2> It's hosted on the wonderful Codeberg Pages</a>.</p> Conclusion</a></h2> <pun></code> I'm terrible at making conclusions. </pun></code></p>

Conclusion</a></h2>
If you have the attention span to read all the way until here, you’re doing just fine. Your brain hasn’t completely decomposed yet and your soul is still within you and has not been sucked away into the depths of some algorithmic feed.</p>
Thanks for reading!</p>

Further reading</a></h2>

https://en.wikipedia.org/wiki/Compose_key</a></li>
https://github.com/kragen/xcompose</a></li>
http://canonical.org/~kragen/setting-up-keyboard.html</a></li> </ul>

`Styling</a></h2> I'm using SCSS</a> to write the stylesheets. The font is IBM Plex Mono</a> which I got from xz/fonts</a>.</p> Hosting</a></h2> It's hosted on the wonderful Codeberg Pages</a>.</p> Conclusion</a></h2> <pun></code> I'm terrible at making conclusions. </pun></code></p>`

Hosting</a></h2>
It's hosted on the wonderful Codeberg Pages</a>.</p>

Conclusion</a></h2>
`<pun></code> I'm terrible at making conclusions. </pun></code></p>`

nik's site - Blog posts

Adding autotracking to a security camera

keyd: Layers, macros, and remaps, even on laptop keyboards

Dynamic MathJax typesetting in Svelte

Escaping the Doomscroll

Adding a search engine to Firefox

The Compose Key: Typing alternate characters the easy way

On writing

Setting up Zola for fun and profit. Also, hello world!

nik's site - Blog posts

Adding autotracking to a security camera

A man, a plan, a lan</a></h2> I wanted to do the following things:</p> Make the camera usable without the shitware app.</li> Try to fix/implement autotracking.</li>

keyd: Layers, macros, and remaps, even on laptop keyboards

Dynamic MathJax typesetting in Svelte

Escaping the Doomscroll

Adding a search engine to Firefox

The Compose Key: Typing alternate characters the easy way

Compose key mappings</a></h2> You can create your own compose key mappings by defining them in the ~/.Xcompose</code> file, but I have personally found this to be a bit finicky.</p> Some default mappings that I often use</a></h2> Superscript numbers: Compose</kbd> ^</kbd> 8</kbd> for ⁸</li>

Some default mappings that I often use</a></h2> Superscript numbers: Compose</kbd> ^</kbd> 8</kbd> for ⁸</li>

On writing

Categorising writing</a></h2> I noticed that my writings fall on a gradient, whose two extremes are:</p> Writing purely for myself</li>

Setting up Zola for fun and profit. Also, hello world!