> "You and me could find crazy that people would openly choose to use IDEs built on HTML/CSS/JS"
Why is that crazy? Makes perfect sense to me, especially as the performance of the programming language element improves. Ultimately these IDEs are sure to make use of WebAssembly, which should take away the remaining performance concerns.
It's crazy because, apart from performance concerns (it looks like we love to make our computers slower and slower every 10 years), you're just discarding all the features of the containing desktop OS. Formatted copypasting, usability features, network features, etc etc... you'll have to reimplement them all, solving all the problems that systems developers solved 10 or 20 years ago. The OS will become little more than a very expensive pixel pipe. But that's what people like, because C++ is hard, native widgets are hard to customise, and everyone loves designing interfaces, so that's where we're going.
I anyone ever starts making javascript-optimised CPUs and GPUs, they're going to make billions. At the moment we only have micropython, but who knows...
Which network features do you need for a coders editor? In the case of Atom/VS Code, can't think of a single one that a browser engine doesn't already provide. Not going to need things like AD-integration, etc...
> "The OS will become little more than a very expensive pixel pipe."
If that's what people want, then so be it. I see no problem with simplifying the OS, most of them are already too bloated.
> " But that's what people like, because C++ is hard, native widgets are hard to customise, and everyone loves designing interfaces, so that's where we're going."
The main advantage is the cross platform compatibility. If certain OS vendors didn't make it hard to build apps that utilised a common base then there'd be much less drive to produce web apps. It has very little to do with the complexity of C++.
> "I anyone ever starts making javascript-optimised CPUs and GPUs, they're going to make billions."
As I've said a number of times now, the final target with web apps won't be JS, it'll be WebAssembly.
Not to mention death of accessibility features - those HTML apps are effectively unusable for disabled users because they don't implement accessibilty features available in native UIs.
Sure, but with the end benefit being a universal UI toolkit.
Also, 'expensively' is debatable, I'm sure it'd be possible to have reusable accessibility components, wouldn't necessarily have to reinvent the wheel for each new web app.
So we finally have a free choice of OS.
It's not the full story to complain about people solving the problems that the OS side had already solved decades ago. The big new thing is that they are doing so in a platform agnostic way. That I have nine different virtual machines installed on four different operating systems and the same code base can run on all of them smoothly.
Except that's not true: Webkit, Gecko, Trident... they are all different "OSs" you're writing for, you just wave them away by shipping the OS with the application. You could do the same by shipping a virtualised image running a stripped-down Linux configured to run only one application. One of these solutions is now socially acceptable, but both manage to completely discard everything the desktop OS achieved in 30 years.
Not really, you still a browser that implement all this stuff. The only difference is Open tech vs proprietary.
Open technologies are obviously a good thing. Writing a software as complex as Photoshop for instance with the exact same features with HTML/CSS and JavaScript isn't going to fly and be usable for someone who has to work 10 hours a day on it. The performance issues will be significant. It's not a big deal for an text editor though I still can't open a 5mb log file in Atom for some reasons. No problem with Sublime Text 2 or Vim. Why is this ?
My point wasn't that they are a silver bullet. My point was simply to say that while something is obviously lost (performance, native integration, etc...), something else is gained. And that is massive portability.
Why is that crazy? Makes perfect sense to me, especially as the performance of the programming language element improves. Ultimately these IDEs are sure to make use of WebAssembly, which should take away the remaining performance concerns.