Electron is a large step backwards in usability and software quality. Spotify has its own context menus which don't support arrow keys or type select. GitHub Desktop has a "Select All" which doesn't actually Select All, only what's been paged in. VSCode doesn't confirm key equivalents by highlighting the menu bar. All of them blink the cursor even without typing focus. All of them steal focus at launch.<p>Our UI vocabulary is shrinking. Undo support used to be table stakes, but now I'm surprised when it works. The web is not only eating the desktop: it's eating what it means to be an expert computer user.
The webification of Desktop app GUI’s is the result of a total disregard for usability .. from a user’s point-of-view (not a developer’s point-of-view of what users >should< want). Most young devs I speak to stare blankly at me when I enquire about GUI standards and usability best-practices. They are now foreign concepts to a lot of developers.<p>As a user, I don’t care about the dev’s challenge of supporting multiple OS’s. These best experience comes when I load up a new application and can intuitively navigate it without the mental overload required trying to decipher a GUI layout which doesn’t adhere to any standard.<p>Web frameworks are great, but have their limitations.<p>The blame is not so much with the new generations of devs, but with the likes of Microsoft and Apple who have ditched standards and are encouraging the Desktop to turn into the wild-west. Or something like that.
The main advantage of writing a program instead of an app is that you can utilize all of the power of the underlying operating system to give the user an experience that is fast, user friendly, and efficient.<p>If everyone is willing to throw that away, say goodbye to general purpose computing, and say welcome to your new App Store Overlords.<p>The last GUI program I wrote in Lazarus worked out to a 24 megabyte executable, because I left the debug information in. It used MySQL and did a lot of computing with gear information. I didn't have to sign the code, or ask ANYONE's permission to deploy it.<p>I'm not giving up general purpose computing, neither should you.
I've been saying a a version of this (the browser will eat the OS) for a few years now, really as soon as I heard about WASM.<p>Once the executives learn that they can deliver binary blobs through the browser, that's the end game for the current distribution model. Piracy will be crippled at the knees and everything will become SaaS.<p>Microsoft has signaled towards this direction when they talked about 10 being the last version of the OS, and the Office suite slowly porting more features to Office 365. Have you seen how aggressive the OneDrive nags have become recently?<p>I fear for a world where PC's more closely resemble Chromebooks than the traditional offline first binary model of today. I can see the benefits this could bring to an end user, but I personally feel the price of freedom and privacy is too high.
I think this is probably true, but I kind of hate it. It's such a staggering waste of computing resources. I don't know, maybe we can get to the point where these apps that include an entire browser are highly optimized but it's still kinda nuts.
I'm early in the tinkering/prototyping phase of a developer tool (not a text editor) that is native AppKit, and it'll stay that way if/when I release it.<p>I actually find myself a lot <i>more</i> productive (i.e. can build a given feature faster) with Xcode+Swift+AppKit than VSCode+Typescript+React+Electron[0], and the near-unanimous disdain for Electron-based apps here gives me a lot of confidence that being a truly native app might actually be its own selling point.<p>[^0]: I learned to program in the early iOS days, spending 3–4 years tinkering around before moving to web dev for the past ~7 years. This is my first project using Swift.
"Web" desktop software adds layers of abstraction and complexity to poorly emulate functionality that already exists in order to decrease the cognitive load of programmers learning paradigms they aren't familiar with. It's less about everything looking like a nail when you only know how to use a hammer, and more about not realizing that there's a better tool for the job. One that is less prone to errors and actually easier to learn to use.<p>That's fine, though. My plan for retirement is to make millions supporting desktop apps that can't be replaced because the "web for desktop" programmers have a ceiling on the level of app complexity they can handle. Line of business apps require 1 to 2 orders of magnitude more effort and so the things written before that paradigm are not economically feasible to replace.<p>If you thought knowing COBOL in 1999 was profitable, just wait 20 years until only a few old-timers know how to maintain the millions of custom WinForms or VB apps that make businesses work.
Why is the post focusing exclusively on desktop?<p>> The last five applications I installed are Electron apps.<p>The last few companies I've worked at made their mobile apps using React Native[1].<p>> In ten years nearly every desktop app that isn't over ten years old will be a progressive web app or a containerized (e.g. Electron) web app.<p>Some of the apps I have installed on my phone are progressive apps, one of those decided to promote their PWA over App Store / Play Store because they didn't want to share revenue with Apple/Google, the revenue issues with mobile stores have been in the news a lot recently, I wouldn't be surprised if this became more common in the future.<p>[1] worth noting I'm not a mobile dev so my own skillset doesn't have anything to do with how those companies make their apps.
I think this take is right in the sense that most consumer-oriented shitware and business CRUD programs are going to be written as web apps from here on out, but I don't really buy that it will completely supplant native software. If you're reading this site, you're probably in tune with the latest hype cycles, and see the world in that lens.<p>In the same vein as the Dark Matter Developer post that occasionally pops up on here, most of the world's software is Dark Matter Software. Mountains of native software quietly runs the world, completely oblivious to the churn we're discussing. Replacing their functionality with a webapp is either impossible or just not worth it.
I dunno. The trouble with web apps is that they keep changing, usually to the detriment of the user. The first one is always free. Then they start charging. Or they add ads. Or they use your info for marketing.
Should have a [2017] appended to the title. When I first saw the title I thought "You're a bit late", but when I clicked on the article and saw it was from Dec. 2017 I thought this is just around the time I felt like that tipping point was reached.
How does security factor into this? Seems like we keep exposing more and more data to networks while at the same time collecting more and more data on everyone. It doesn't seem like a great combination for those who have to protect this data nor for those who will be the most harmed if it gets out (the folks the data is about, not the companies that leaked it).
I’m not sure one will eat the other. The two will converge. The technologies typical of desktop apps like GPU acceleration, type safe languages, AoT compilation, multithreading etc - all of those things have rather recently moved from being somewhat or entirely desktop exclusive, to being on the web.<p>With novel means of deploying sandboxed apps the line will continue to blur.
I always thought that deploying web apps as desktop apps is a great thing to do. As in, not bunch it up into electron, but using shell scripting to start a local only web server and send a command to the OS to open a localhost page.<p>The advantages are that you don't need bloated software to do stuff, you have the full power of the OS, because you have a local webserver running, you have access to the file system, etc, you have the choice to use any programming language and framework (not be limited to JS), etc.<p>I keep thinking why people don't go this way. Are there any glaring mistakes to this approach I am unable to see?
When there was a single platform (Windows), native made sense as either way it was one codebase. You now need to support at least two desktop platforms and some computers now only support web stuff (Chromebooks).<p>The only argument for native is if resources are such that make a web app impractical for most people (high performance GPU intensive stiff). Otherwise, sure, Electron isn't that efficient and takes up relatively a lot of space, but we are at the point where space and RAM are abundant, so it is not a limiting factor for most people.
> I took the post offline after about a week because I received among other things DDOS attacks and frothing at the mouth e-mails including one sent from a throwaway e-mail that actually made threats of physical violence against myself and my family.
What is wrong with people? It’s not like he wants crappy Electron w/Javascript to replace desktop with all its flaws! He acknowledges the main issues for webapps and proposes solutions.<p>OP, if you’re reading this, I’m sorry you had to go through that.<p>EDIT: misread. Still doesn’t justify any of the hate.
Even though electron apps are strange and quirky compared to native apps, I am thankful they are being made.<p>The other option is what Google is doing. No native apps at all. Have a chrome tab open for everything. Chat, Gmail, calender, meet, YouTube music. Evrything in its own tab, no apps for any platform. Fuck that.
I don't know how well thought this article is, due to a number of use-cases where desktop apps dominate (high performance, GPU, extreme latency sensitivity), and I don't see that changing. I think we are more likely to see some stratification between different types of apps.
I really wish flutter to succeed on Desktop.
It checks most of the points.
1. Modern reactive API like React
2. Better rendering performance than electron
3. Interoperable API between web/mobile/desktop
very likely, if you take a sort of long historical view then it's I think intuitive to predict that the ark of computing bends towards greater and greater connectivity, and the web is the basic infrastructure for that. Hardware and operating systems just become details and everyone hooks into the matrix pretty much.<p>The piece is from 2017 but it's pretty prescient when it recommends web based interfaces as a way for linux to leapfrog into that world. Reminds me of Flutter being adopted by Ubuntu recently which does very well as a web/desktop/mobile platform.
> In ten years nearly every desktop app that isn't over ten years old will be a progressive web app or a containerized (e.g. Electron) web app.<p>What proportion of desktop apps are computer games?
The usual retort is "people use Electron because they only know web technologies".<p>No, it's not only that.<p>I've been programming GUI apps for 30 years, going through TurboVision, Delphi, MFC, WinForms, wxWidgets, Qt, WPF, native Android/iPhone, and probably 2 more I forgot about.<p>They all suck terribly when compared to React/Vue/HTML/CSS.<p>Creating a GUI app with web tech it's just so much more productive. It's like comparing the productivity of Python to C++. Sure, you can't write an audio editor in Python/Electron, but for 90% of GUI apps it makes no sense to use native frameworks. They are as productive and expressive as writing in assembly.
Since the gui-fication of everything, nothing has made making a gui easier than web apps. The reason I like it is because of the cross platform capabilities. I don't have to worry about compiling for different platforms. I don't need to worry about platform specific issues. It's the ultimate form of portable code honestly nowadays.
The benefits of web based apps in the majority of cases I would say far outweigh any other issues.<p>Right now I am working on a side project in Flutter, it is in the very early stages but the ability to target multiple platforms, web, multiple desktop environments, multiple mobile platforms from a fairly singular code base is rather interesting. This makes good commercial sense, in practical thinking speeds up development and release. Yep issues may arise.<p>I would not hesitate to say that for many users the average users that web based apps, either browser based or electron or whatever are suitable in a lot of cases. There are some cases where this does not make sense currently, like computational (although), video/audio editing, etc. However look at tools like Google Colab where machine learning is being pushed server side because of the power of 3rd party infrastructure.<p>The web just makes sense in a lot of cases. Ease of development, maintainability, cost, UI, accessibility (availability).