More from Marco.org
Ten years ago, Apple’s Phil Schiller surprised Apple enthusiasts and developers by walking out on stage at John Gruber’s The Talk Show Live WWDC event and giving an open, human, honest interview to a somewhat jaded community. I wrote this in response: Both Apple and Phil Schiller himself took a huge risk in doing this. That they agreed at all is a noteworthy gift to this community of long-time enthusiasts, many of whom have felt under-appreciated as the company has grown. […] Phil’s appearance on the show was warm, genuine, informative, and entertaining. It was human. And humanizing the company and its decisions, especially to developers — remember, developer relations is all under Phil — might be worth the PR risk. This started a ten-year run of interviews by Apple executives on The Talk Show every year at WWDC that proved to be great, surprisingly safe PR for Apple. No executive ever said something they shouldn’t have (they’re pros), no sensational or negative news stories ever resulted from them, and Apple’s enthusiastic fans and developers felt seen, heard, and appreciated. * * * For unspecified reasons, Apple has declined to participate this year, ending what had become a beloved tradition in our community — and I can’t help but suspect that it won’t come back. (A lot has changed in the meantime.) Maybe Apple has good reasons. Maybe not. We’ll see what their WWDC PR strategy looks like in a couple of weeks. In the absence of any other information, it’s easy to assume that Apple no longer wants its executives to be interviewed in a human, unscripted, unedited context that may contain hard questions, and that Apple no longer feels it necessary to show their appreciation to our community and developers in this way. I hope that’s either not the case, or it doesn’t stay the case for long. This will be the first WWDC I’m not attending since 2009 (excluding the remote 2020 one, of course). Given my realizations about my relationship with Apple and how they view developers, I’ve decided that it’s best for me to take a break this year, gain some perspective, and decide what my future relationship should look like. Maybe Apple’s leaders are doing that, too.
Today, on the tenth anniversary of Overcast 1.0, I’m happy to launch a complete rewrite and redesign of most of the iOS app, built to carry Overcast into the next decade — and hopefully beyond. Like podcasts better than blog posts? Listen to ATP #596 for more! What’s new Much faster, more responsive, more reliable, and more accessible. Modern design, optimized for easily-reached controls on today’s phone sizes. Improvements throughout, such as undoing large seeks, new playlist-priority options, easier navigation, and more. What’s not Most features. Overcast is still Overcast! The audio engine. It’s the best part of Overcast, and still leads the industry in sound quality, silence skipping, and volume normalization. (More soon!) The business. I’m still a one-person operation, with no funding or external ownership, serving only my customers. My principles. I always want to make the best podcast app, and I’ll never disrespect your time, attention, or privacy. What’s gone Streaming. Most big podcasts now use dynamic ad insertion, which causes bugs and problems for streaming playback.1 Downloading episodes completely before they begin playback is much more reliable. Tapping a non-downloading episode will now open the playback screen, download it, then start playback. It works similarly to the way streaming did before, but playback begins after the download completes, not after a portion of it is buffered. On today’s fast networks, this usually only takes a few extra seconds. And in the near future, I’ll be adding smarter options and more control over selective downloading of episodes to further improve the experience for people who don’t automatically download every episode. What’s next The last few missing features from the old app, such as Shortcuts support, storage management, and OPML. These are absent now, but will return soon. More options for downloading and deleting episodes. Upgrading the Apple Watch app to the new, faster sync engine. (The Watch app is currently unchanged from the previous one.) And, of course, more features, including some of your most-requested features over the last decade. Getting this rewrite out the door was a monumental task. Thank you for your patience as I work through this list! Why? Most of Overcast’s core code was 10 years old, which made it cumbersome or impossible to easily move with the times, adopt new iOS functionality, or add new features, especially as one person. That’s why there haven’t been many new features or changes in years. You saw it, and I saw it. I wasn’t able to serve my customers as well as I wanted. For Overcast to have a future, it needed a modern foundation for its second decade. I’ve spent the past 18 months rebuilding most of the app with Swift, SwiftUI, Blackbird, and modern Swift concurrency. Now, development is rapidly accelerating. I’m more responsive, iterating more quickly, and ultimately making the app much better. Thank you all so much for the first decade of Overcast. Here’s to the next one. Dynamic ad insertion (DAI) splices ads into each download, and no two downloads are guaranteed to have the same number or duration of ads. So, for example, if the first half of an episode downloads, then the download fails, and it downloads the second half with another request, the combined audio may jump forward or back at the halfway mark, losing or repeating content. ↩︎
Overcast’s latest update (2022.2) brings the largest redesign in its nearly-eight-year history, plus many of the most frequently requested features and lots of under-the-hood improvements. I’m pretty proud of this one. For this first and largest phase of the redesign, I focused on the home screen, playlist screen, typography, and spacing. (I plan to revamp the now-playing and individual-podcast screens in a later update.) The home screen is radically different: Home screen, before (left) and after (right). Playlists now have strong visual identities for nicer and easier navigation. Each playlist has a customizable color, and a custom icon can be selected from over 3,000 SF Symbols to match modern iOS design and the other icons within Overcast. And playlists can be manually reordered with drag-and-drop. Recently played and newly published episodes can now be displayed on the home screen for quick access, much like the widget and CarPlay experience. Podcasts can now be pinned to the top of the home-screen list. Pinned podcasts can also be manually reordered with drag-and-drop. I’ve also rethought the old stacked “Podcasts” and “Played Podcasts” sections to better match people’s needs and expectations. Now, the toggle atop the podcast list switches between three modes: podcasts with current episodes, all followed podcasts, and inactive podcasts (those that you don’t follow and therefore won’t get any more episodes from, or haven’t posted a new episode in a long time). The playlist screen’s structure remains mostly the same, while refining the design for the modern era: Playlist screen, before (left) and after (right). Here, it’s more apparent that I’ve replaced the system San Francisco font with an alternate variant, San Francisco Rounded, to increase legibility and better match the personality of the app. I’ve also added highly demanded features: By far, Overcast’s most-requested feature is a Mark as Played feature. That’s now available as a checkmark button on episode rows, as well as a left-side swipe action. The second-most-requested feature is a way to view all starred episodes. Special playlists for Starred, Downloaded, and In Progress can now be created. The light and dark themes now each have a customizable tint color from the modern iOS UI-color palette, including these favorites from beta testers: And throughout the app, I’ve made tons of tweaks and bug fixes, including: Notifications and background downloads are now much more reliable. Episode downloads can now be individually deleted or re-downloaded. Links can now be opened in Safari. (under Nitpicky Details) Performance is now significantly better with very large playlists and collections. Fixed bugs with episode-duration detection, CarPlay lists, Mac-app sharing, and much more. So much is better in this update that I can’t even remember it all. Thank you so much to everyone who helped me beta-test this massive update. As always, Overcast is free in the App Store. Go get it!
After the dust settles from the developer class-action settlement, the South Korean law, the JFTC announcement, and the Apple v. Epic decision, I think the most likely long-term outcome isn’t very different from the status quo — and that’s a good thing. Allowing external purchases Here’s what I think we’ll end up with: Apple will still require apps to use their IAP system for any qualifying purchases that occur in the apps themselves. All app types will be allowed to link out to a browser for other purchase methods. Most apps will be required to also offer IAP side-by-side with any external methods.1 Only “Reader apps” will be exempt from this requirement.2 Apple will have many rules regarding the display, descriptions, and behavior of external purchases, many of which will be unpublished and ever-changing. App Review will be extremely harsh, inconsistent, capricious, petty, and punitive with their enforcement.3 Apple won’t require price-matching between IAP and external purchases. These few but important corrections reduce Apple’s worst behavior and should relieve most regulatory pressure. The result won’t look much different than the status quo: Most big media apps (qualifying as “reader” apps) won’t offer IAP, but will finally be allowed to link to their websites from their apps and offer purchases there. Many games will offer both IAP and external purchases, with the external choice offering a discount, bonus gems, extra loot boxes, or other manipulative tricks to optimize the profitability of casino games for children (commissions from which have been the largest portion of Apple’s “services revenue” to date). Most importantly, many products, services, and business models will become possible that previously weren’t, leading to more apps, more competition, and more money going to more places. External purchase methods will evolve to be almost as convenient as IAP (especially if Apple Pay is permitted in this context), and payment processors will reduce the burden of manual credit-card entry with shared credentials available across multiple apps. The payment-fraud doomsday scenarios argued by Apple and many fans mostly won’t happen, in part because App Review will prevent most obvious cases, but also because parents don’t typically offer their credit cards to untrustworthy children; and for buyers of all ages, most credit cards themselves provide stronger fraud prevention and easier recourse from unwanted charges than the App Store ever has. No side-loading I don’t expect side-loading or alternative app stores to become possible, and I’m relieved, because that is not a future I want for iOS. When evaluating such ideas, I merely ask myself: “What would Facebook do?” Facebook owns four of the top ten apps in the world. If side-loading became possible, Facebook could remove Instagram, WhatsApp, the Facebook app, and Messenger from Apple’s App Store, requiring customers to install these extremely popular apps directly from Facebook via side-loading. And everyone would. Most people use a Facebook-owned app not because it’s a good app, but because it’s a means to an important end in their life. Social pressure, family pressure, and network lock-in prevent most users from seeking meaningful alternatives. People would jump through a few hoops if they had to. Facebook would soon have apps that bypassed App Review installed on the majority of iPhones in the world. Technical limitations of the OS would prevent the most egregious abuses, but there’s a lot they could still do. We don’t need to do much imagining — they already have attempted multiple hacks, workarounds, privacy invasions, and other unscrupulous and technically invasive behavior with their apps over time to surveil user behavior outside of their app and stay running longer in the background than users intend or expect. The OS could evolve over time to reduce some of these vulnerabilities, but technical measures alone cannot address all of them. Without the threat of App Review to keep them in check, Facebook’s apps would become even more monstrous than they already are. As a user and a fan of iOS, I don’t want any part of that. No alternative app stores Alternative app stores would be even worse. Rather than offering individual apps via side-loading, Facebook could offer just one: The Facebook App Store. Instagram, WhatsApp, the Facebook app, and Messenger could all be available exclusively there. The majority of iOS users in the world would soon install it, and Facebook would start using leverage in other areas — apps’ social accounts, stats packages, app-install ads, ad-attribution requirements — to heavily incentivize (and likely strong-arm) a huge number of developers to offer their apps in the Facebook App Store, likely in addition to Apple’s. Maybe I’d be required to add the Facebook SDK to my app in order to be in their store, which they would then use to surveil my users. Maybe I’d need to buy app-install ads to show up in search there at all. Maybe I’d need to pay Facebook to “promote” each app update to reach more than a tiny percentage of my existing customers. And Facebook wouldn’t even be the only app store likely to become a large player on iOS. Amazon would almost certainly bring their garbage “Appstore” to iOS, but at least that one probably wouldn’t go anywhere. Maybe Google would bring the Play Store to iOS and offer a unified SDK to develop a single codebase for iOS and Android, effectively making every app feel like an Android app and further marginalizing native apps when they’re already hurting. Media conglomerates that own many big-name properties, like Disney, might each have their own app stores for their high-profile apps. Running your own store means you can promote all of your own apps as much as you want. What giant corporation would resist? Don’t forget games! Epic and Steam would come to iOS with their own game stores. Maybe Microsoft and Nintendo, too. Maybe you’d need to install seven different app stores on your iPhone just to get the apps and games you already use — and all without App Review to keep them in check. Most developers would probably need to start submitting our apps to multiple app stores, each with its own rules, metadata, technical requirements, capabilities, approval delays, payment processing, stats, crash reports, ads, promotion methods, and user reviews. As a user, a multiple-app-store world sounds like an annoying mess; as a developer, it terrifies me. Apple’s App Store is the devil we know. The most viable alternatives that would crop up would be far worse. Course correction The way Apple runs its business isn’t perfect, but it’s also not a democracy. I loved this part of Judge Yvonne Gonzalez Rogers’ decision in Apple v. Epic, as quoted by Ben Thompson’s excellent article that you should read: Apple has not offered any justification for the actions other than to argue entitlement. Where its actions harm competition and result in supracompetitive pricing and profits, Apple is wrong. I interpret “entitlement” without a negative connotation here — Apple is entitled to run their platform mostly as they wish, with governmental interference only warranted to fix market-scale issues that harm large segments of commerce or society. As a developer, I’d love to see more changes to Apple’s control over iOS. But it’s hard to make larger changes without potentially harming much of what makes iOS great for both users and developers. Judge Gonzalez Rogers got it right: we needed a minor course correction to address the most egregiously anticompetitive behavior, but most of the way Apple runs iOS is best left to Apple. If the South Korean law holds, IAP may not be required — but only in South Korea. With this exception, I expect the rest of these rules to be enforced the same way globally. ↩︎ Apple defines “reader” apps as “[allowing] a user to access previously purchased content or content subscriptions (specifically: magazines, newspapers, books, audio, music, and video).” This includes many apps that Apple’s services compete with, such as Netflix, Spotify, and Kindle, that raise anticompetitive concerns among regulators and legislators when forced to give Apple 30%. ↩︎ App Review has higher-level queues for managerial review of controversial rules or edge cases, typically identifiable from the outside by an app stuck with “In Review” status for days or weeks, and often ending in a phone call from “Bill”. I’d expect any app offering external purchases to have a very high chance of being escalated to a slower, more pain-in-the-ass review process, possibly causing it not to be worthwhile for most small developers to deal with. I have no plans to add external purchases to Overcast for multiple reasons, including this — but mostly because, for my purposes, I’m satisfied with Apple’s IAP system. ↩︎
More in programming
Here are a few tangentially-related ideas vaguely near the theme of comparison operators. comparison style clamp style clamp is median clamp in range range style style clash? comparison style Some languages such as BCPL, Icon, Python have chained comparison operators, like if min <= x <= max: ... In languages without chained comparison, I like to write comparisons as if they were chained, like, if min <= x && x <= max { // ... } A rule of thumb is to prefer less than (or equal) operators and avoid greater than. In a sequence of comparisons, order values from (expected) least to greatest. clamp style The clamp() function ensures a value is between some min and max, def clamp(min, x, max): if x < min: return min if max < x: return max return x I like to order its arguments matching the expected order of the values, following my rule of thumb for comparisons. (I used that flavour of clamp() in my article about GCRA.) But I seem to be unusual in this preference, based on a few examples I have seen recently. clamp is median Last month, Fabian Giesen pointed out a way to resolve this difference of opinion: A function that returns the median of three values is equivalent to a clamp() function that doesn’t care about the order of its arguments. This version is written so that it returns NaN if any of its arguments is NaN. (When an argument is NaN, both of its comparisons will be false.) fn med3(a: f64, b: f64, c: f64) -> f64 { match (a <= b, b <= c, c <= a) { (false, false, false) => f64::NAN, (false, false, true) => b, // a > b > c (false, true, false) => a, // c > a > b (false, true, true) => c, // b <= c <= a (true, false, false) => c, // b > c > a (true, false, true) => a, // c <= a <= b (true, true, false) => b, // a <= b <= c (true, true, true) => b, // a == b == c } } When two of its arguments are constant, med3() should compile to the same code as a simple clamp(); but med3()’s misuse-resistance comes at a small cost when the arguments are not known at compile time. clamp in range If your language has proper range types, there is a nicer way to make clamp() resistant to misuse: fn clamp(x: f64, r: RangeInclusive<f64>) -> f64 { let (&min,&max) = (r.start(), r.end()); if x < min { return min } if max < x { return max } return x; } let x = clamp(x, MIN..=MAX); range style For a long time I have been fond of the idea of a simple counting for loop that matches the syntax of chained comparisons, like for min <= x <= max: ... By itself this is silly: too cute and too ad-hoc. I’m also dissatisfied with the range or slice syntax in basically every programming language I’ve seen. I thought it might be nice if the cute comparison and iteration syntaxes were aspects of a more generally useful range syntax, but I couldn’t make it work. Until recently when I realised I could make use of prefix or mixfix syntax, instead of confining myself to infix. So now my fantasy pet range syntax looks like >= min < max // half-open >= min <= max // inclusive And you might use it in a pattern match if x is >= min < max { // ... } Or as an iterator for x in >= min < max { // ... } Or to take a slice xs[>= min < max] style clash? It’s kind of ironic that these range examples don’t follow the left-to-right, lesser-to-greater rule of thumb that this post started off with. (x is not lexically between min and max!) But that rule of thumb is really intended for languages such as C that don’t have ranges. Careful stylistic conventions can help to avoid mistakes in nontrivial conditional expressions. It’s much better if language and library features reduce the need for nontrivial conditions and catch mistakes automatically.
After shipping my work transforming HTML with Netlify’s edge functions I realized I have a little bug: the order of the icons specified in the URL doesn’t match the order in which they are displayed on screen. Why’s this happening? I have a bunch of links in my HTML document, like this: <icon-list> <a href="/1/">…</a> <a href="/2/">…</a> <a href="/3/">…</a> <!-- 2000+ more --> </icon-list> I use html-rewriter in my edge function to strip out the HTML for icons not specified in the URL. So for a request to: /lookup?id=1&id=2 My HTML will be transformed like so: <icon-list> <!-- Parser keeps these two --> <a href="/1/">…</a> <a href="/2/">…</a> <!-- But removes this one --> <a href="/3/">…</a> </icon-list> Resulting in less HTML over the wire to the client. But what about the order of the IDs in the URL? What if the request is to: /lookup?id=2&id=1 Instead of: /lookup?id=1&id=2 In the source HTML document containing all the icons, they’re marked up in reverse chronological order. But the request for this page may specify a different order for icons in the URL. So how do I rewrite the HTML to match the URL’s ordering? The problem is that html-rewriter doesn’t give me a fully-parsed DOM to work with. I can’t do things like “move this node to the top” or “move this node to position x”. With html-rewriter, you only “see” each element as it streams past. Once it passes by, your chance at modifying it is gone. (It seems that’s just the way these edge function tools are designed to work, keeps them lean and performant and I can’t shoot myself in the foot). So how do I change the icon’s display order to match what’s in the URL if I can’t modify the order of the elements in the HTML? CSS to the rescue! Because my markup is just a bunch of <a> tags inside a custom element and I’m using CSS grid for layout, I can use the order property in CSS! All the IDs are in the URL, and their position as parameters has meaning, so I assign their ordering to each element as it passes by html-rewriter. Here’s some pseudo code: // Get all the IDs in the URL const ids = url.searchParams.getAll("id"); // Select all the icons in the HTML rewriter.on("icon-list a", { element: (element) => { // Get the ID const id = element.getAttribute('id'); // If it's in our list, set it's order // position from the URL if (ids.includes(id)) { const order = ids.indexOf(id); element.setAttribute( "style", `order: ${order}` ); // Otherwise, remove it } else { element.remove(); } }, }); Boom! I didn’t have to change the order in the source HTML document, but I can still get the displaying ordering to match what’s in the URL. I love shifty little workarounds like this! Email · Mastodon · Bluesky
I realize that for all I've talked about Logic for Programmers in this newsletter, I never once explained basic logical quantifiers. They're both simple and incredibly useful, so let's do that this week! Sets and quantifiers A set is a collection of unordered, unique elements. {1, 2, 3, …} is a set, as are "every programming language", "every programming language's Wikipedia page", and "every function ever defined in any programming language's standard library". You can put whatever you want in a set, with some very specific limitations to avoid certain paradoxes.2 Once we have a set, we can ask "is something true for all elements of the set" and "is something true for at least one element of the set?" IE, is it true that every programming language has a set collection type in the core language? We would write it like this: # all of them all l in ProgrammingLanguages: HasSetType(l) # at least one some l in ProgrammingLanguages: HasSetType(l) This is the notation I use in the book because it's easy to read, type, and search for. Mathematicians historically had a few different formats; the one I grew up with was ∀x ∈ set: P(x) to mean all x in set, and ∃ to mean some. I use these when writing for just myself, but find them confusing to programmers when communicating. "All" and "some" are respectively referred to as "universal" and "existential" quantifiers. Some cool properties We can simplify expressions with quantifiers, in the same way that we can simplify !(x && y) to !x || !y. First of all, quantifiers are commutative with themselves. some x: some y: P(x,y) is the same as some y: some x: P(x, y). For this reason we can write some x, y: P(x,y) as shorthand. We can even do this when quantifying over different sets, writing some x, x' in X, y in Y instead of some x, x' in X: some y in Y. We can not do this with "alternating quantifiers": all p in Person: some m in Person: Mother(m, p) says that every person has a mother. some m in Person: all p in Person: Mother(m, p) says that someone is every person's mother. Second, existentials distribute over || while universals distribute over &&. "There is some url which returns a 403 or 404" is the same as "there is some url which returns a 403 or some url that returns a 404", and "all PRs pass the linter and the test suites" is the same as "all PRs pass the linter and all PRs pass the test suites". Finally, some and all are duals: some x: P(x) == !(all x: !P(x)), and vice-versa. Intuitively: if some file is malicious, it's not true that all files are benign. All these rules together mean we can manipulate quantifiers almost as easily as we can manipulate regular booleans, putting them in whatever form is easiest to use in programming. Speaking of which, how do we use this in in programming? How we use this in programming First of all, people clearly have a need for directly using quantifiers in code. If we have something of the form: for x in list: if P(x): return true return false That's just some x in list: P(x). And this is a prevalent pattern, as you can see by using GitHub code search. It finds over 500k examples of this pattern in Python alone! That can be simplified via using the language's built-in quantifiers: the Python would be any(P(x) for x in list). (Note this is not quantifying over sets but iterables. But the idea translates cleanly enough.) More generally, quantifiers are a key way we express higher-level properties of software. What does it mean for a list to be sorted in ascending order? That all i, j in 0..<len(l): if i < j then l[i] <= l[j]. When should a ratchet test fail? When some f in functions - exceptions: Uses(f, bad_function). Should the image classifier work upside down? all i in images: classify(i) == classify(rotate(i, 180)). These are the properties we verify with tests and types and MISU and whatnot;1 it helps to be able to make them explicit! One cool use case that'll be in the book's next version: database invariants are universal statements over the set of all records, like all a in accounts: a.balance > 0. That's enforceable with a CHECK constraint. But what about something like all i, i' in intervals: NoOverlap(i, i')? That isn't covered by CHECK, since it spans two rows. Quantifier duality to the rescue! The invariant is equivalent to !(some i, i' in intervals: Overlap(i, i')), so is preserved if the query SELECT COUNT(*) FROM intervals CROSS JOIN intervals … returns 0 rows. This means we can test it via a database trigger.3 There are a lot more use cases for quantifiers, but this is enough to introduce the ideas! Next week's the one year anniversary of the book entering early access, so I'll be writing a bit about that experience and how the book changed. It's crazy how crude v0.1 was compared to the current version. MISU ("make illegal states unrepresentable") means using data representations that rule out invalid values. For example, if you have a location -> Optional(item) lookup and want to make sure that each item is in exactly one location, consider instead changing the map to item -> location. This is a means of implementing the property all i in item, l, l' in location: if ItemIn(i, l) && l != l' then !ItemIn(i, l'). ↩ Specifically, a set can't be an element of itself, which rules out constructing things like "the set of all sets" or "the set of sets that don't contain themselves". ↩ Though note that when you're inserting or updating an interval, you already have that row's fields in the trigger's NEW keyword. So you can just query !(some i in intervals: Overlap(new, i')), which is more efficient. ↩
Email is your most important online account, so keep it clean.
Kubernetes is not exactly the most fun piece of technology around. Learning it isn’t easy, and learning the surrounding ecosystem is even harder. Even those who have managed to tame it are still afraid of getting paged by an ETCD cluster corruption, a Kubelet certificate expiration, or the DNS breaking down (and somehow, it’s always the DNS). Samuel Sianipar If you’re like me, the thought of making your own orchestrator has crossed your mind a few times. The result would, of course, be a magical piece of technology that is both simple to learn and wouldn’t break down every weekend. Sadly, the task seems daunting. Kubernetes is a multi-million lines of code project which has been worked on for more than a decade. The good thing is someone wrote a book that can serve as a good starting point to explore the idea of building our own container orchestrator. This book is named “Build an Orchestrator in Go”, written by Tim Boring, published by Manning. The tasks The basic unit of our container orchestrator is called a “task”. A task represents a single container. It contains configuration data, like the container’s name, image and exposed ports. Most importantly, it indicates the container state, and so acts as a state machine. The state of a task can be Pending, Scheduled, Running, Completed or Failed. Each task will need to interact with a container runtime, through a client. In the book, we use Docker (aka Moby). The client will get its configuration from the task and then proceed to pull the image, create the container and start it. When it is time to finish the task, it will stop the container and remove it. The workers Above the task, we have workers. Each machine in the cluster runs a worker. Workers expose an API through which they receive commands. Those commands are added to a queue to be processed asynchronously. When the queue gets processed, the worker will start or stop tasks using the container client. In addition to exposing the ability to start and stop tasks, the worker must be able to list all the tasks running on it. This demands keeping a task database in the worker’s memory and updating it every time a task change’s state. The worker also needs to be able to provide information about its resources, like the available CPU and memory. The book suggests reading the /proc Linux file system using goprocinfo, but since I use a Mac, I used gopsutil. The manager On top of our cluster of workers, we have the manager. The manager also exposes an API, which allows us to start, stop, and list tasks on the cluster. Every time we want to create a new task, the manager will call a scheduler component. The scheduler has to list the workers that can accept more tasks, assign them a score by suitability and return the best one. When this is done, the manager will send the work to be done using the worker’s API. In the book, the author also suggests that the manager component should keep track of every tasks state by performing regular health checks. Health checks typically consist of querying an HTTP endpoint (i.e. /ready) and checking if it returns 200. In case a health check fails, the manager asks the worker to restart the task. I’m not sure if I agree with this idea. This could lead to the manager and worker having differing opinions about a task state. It will also cause scaling issues: the manager workload will have to grow linearly as we add tasks, and not just when we add workers. As far as I know, in Kubernetes, Kubelet (the equivalent of the worker here) is responsible for performing health checks. The CLI The last part of the project is to create a CLI to make sure our new orchestrator can be used without having to resort to firing up curl. The CLI needs to implement the following features: start a worker start a manager run a task in the cluster stop a task get the task status get the worker node status Using cobra makes this part fairly straightforward. It lets you create very modern feeling command-line apps, with properly formatted help commands and easy argument parsing. Once this is done, we almost have a fully functional orchestrator. We just need to add authentication. And maybe some kind of DaemonSet implementation would be nice. And a way to handle mounting volumes…