But isn't the bloat of electron which enables the greatest benefit of VS Code? By which I mean the plugins with their wide range of abilities and UI-integrations.
> It's like sublime text, but it's open source and has first class LSP support.
Sublime has LSP since a while now.
Maybe for some users. I'm sparing with extensions because I find them hard to vet; I checked and I'm only using clangd and prettier right now. But I don't think Electron is required in order to have rich extensions - just see Vim/Neovim, Emacs, or any IDE.
> Sublime has LSP since a while now.
My bad then. I thought sublime's LSP support came from a third party package.
I don't know about IDEs, but I wouldn't put vim & Emacs on the same level as VS Code in terms of interfaces. They go much further than your average text-editor and also grow over time, but at the end they are still text-based interfaces with some fluff. Though, It of course depends strongly on your requirements. For most developers, text with fluff is more than enough.
But in any case, I try to avoid bespoke custom editor setups unless there's a situation that really warrants it.
Depends on your definition of "whole". Those modes work by removing features from the webstack, or by borrowing from external GUIs. Emacs browser w3m for example, is AFAIK unable to handle JavaScript. Electron-apps on the other side offer you whole webstack, including all available components, widgets, libs, frameworks, or whatever. So it's very easy to take some webapp or component and integrate into your workflows. Apps like Emacs are unable of doing this, and thus constantly on a game of fetching up on popular features and services.
It's one of the things I have about "real" IDEs with a vim mode. The one in vscode is pretty good, it's pretty rare I break it when just editing code (mostly '.' not doing what I expect). But everything else doesn't fit the vim model. I have the file manager or something open in the left and I can't reach it with ctrl+h, because it's own weird UI thing not just a split (like it would be in vim).
As a vim user I can reach a sort of uncanny valley where the core editing loop feels "right," but I end up having to fight against all that muscle memory when it comes to using features outside of it.
Every plugin in nvim required to learn some new different bs shortcut to have basic features. Plus, I ran into plenty of crashes and weird error in some plugins - and performance wasn't always great.
The editing is fine, you can get used to buffer editing - it's everything else which is not polished and feels like a pile of half implemented features.
But the principle I'm talking about applies without any plugins at all. If just do :terminal it'll make a split, you can now switch between whatever you were doing and that terminal the same way you would switch between a regular split.
None of the proper IDEs do that. I can't just look at what's on my screen and do "move focus to the thing below where I am". You need to memorize shortcuts for each bloody tab rather than them being context aware, and I hate it.
At the cost of difficulty. Recently I wrote a VSCode extension myself with zero prior knowledge (shameless plug: https://github.com/CrendKing/universal-format-on-type/), in probably one day. All I needed to study is their extension API that I cared about. No need for learning a whole new language or DSL. Also easy to debug since there is no compilation and little environment setup.
There is a good reason even though everyone acknowledge about Electron's performance cost, everyone is still using it.
Plugins are overrated and I try not to use them as much as I can.