My Neovim configuration[0] is currently the best it's been IMO, but I ended up having a few more plugins than I thought I would need.
I am earlier along in my Emacs rebuild, but what is really interesting to me is that there are a lot of improvements in modern versions of Emacs (30+) that dramatically reduce my dependence on third party plug-ins. Eglot has been fantastic for me, as a former long-time lsp-mode user. Completion preview mode is not quite a replacement for corfu, but it's pretty darn good and only going to get better.
So far my list of must have plug-ins for Emacs:
- Magit - Expreg (teeesitter expand region) - Multiple cursors - dape (debugging in coordination with Eglot)
Thinking I will probably end up adding Consult + orderless, too, as that is a search/navigation superpower.
[0] https://github.com/wilkystyle/nvim
edit to add nvim repo link
IMO Neovim OOB experience isn't quite as nice as what Eglot and completion-preview-mode do for the Emacs OOB experience, but I love that every time I check back in on the latest state of either editor, they are sensibly adding really solid functionality to the core that continuously reduces my reliance on third-party plug-ins!
But because Vim-like keys and modes seem to be so much better integrated in all other editors these days (no good Emacs mode in VSCode for example), seems like it won, and it'll soon be time to accept that.
But every time I tried – I never had enough patience to use it until my fingers have the memory, and didn't see the point enough to keep smashing the editor.
So, very interested in reports of actual hardcore Emacs users about the switch.
Vim (and variants) is a language for editing first, and an editor second. Most of vim parts from a text editor perspective is found in some way or others in other editors. It's biggest pro is being centered about the vim language of editing. Once you master that, editing is very fast and you can extend the language easily by using external tools. And there's already a good number of integrations through builtin mechanisms and various plugins.
Emacs is a collections of tools for working with textual interfaces. Where Emacs easily wins is how easy it it is to integrate with anything that plays with text, (which is basically everything). The good part is that every plugin (even builtin one) is a software, a library, and a playground at the same time.
- Vim modes are prolific, but they always fall short in some annoying way (Jupyterlab-vim plug-in is so bad it's almost not even worth considering Vim emulation). I inevitably find myself missing the full power of true Neovim, especially with the live find/replace in the buffer, `:cdo`, and things like `:g//norm` (which could be an entire tutorial in itself!)
- Emacs: I have never met another editor (or even piece of software) that is not only so customizable but so designed to be customized. It's really hard to capture it in words, but it's such a comprehensive ecosystem of tools that all are designed to work together incredibly well and designed in such a way that when you make your customizations you can do it in a way that everything still plays really nicely together. For example, I can use Tramp to make my local, fully-customized Emacs edit on a remote filesystem as though my local setup is there, any arbitrary number of SSH hops away, even sudo'd to root on the target machine. I can further leverage the built-in /docker: protocol to edit with my local editor, on that remote machine, as root. Eglot plays nicely so my local editor uses remote LSP servers. dired-mode plays nicely so that I can browse and modify the filesystem on the remote target with my local editor. Because everything is nicely integrated together, pressing M-! to run a shell command while editing a remote buffer will run the shell command on the remote target in the same directory as that remote buffer. Similarly, M-x shell opens a terminal window on the remote target, in the same directory as the remote file I'm editing, as whatever user I have sudo'd to (if applicable). And it does everything with pure SSH, no remote daemon needed. And then there is the customization. Vim/Neovim score very well on the customization front, but I guess the way I would put it is that the difference in philosophy is that Vim/Neovim (and most other customizable software) are applications for which the developers give you a ton of APIs to affect the behavior of an otherwise closed system. With Emacs, by contrast, nearly all of the system is open. There are compiled C functions at the core, but anything in lisp is there for inspecting and modifying in real time while the system is running. And for all the things that are compiled that you cannot affect, there are further paradigms such as the advice system, which allows you to modify built-in behavior by wrapping existing functionality with things like pre- or post- actions (among other things too numerous to get into in a comment). emacs is built with the philosophy that there should never be an area of the software that you don't fully own, control, and have the ability to inspect and understand.
I guess this turned into a classic Emacs-nerd rant about why Emacs is great, but to answer your question the long way:
There's no Vim like (Neo)vim, and if you are an advanced enough (Neo)vim user, everything else falls short in ways that suck.
And there is simply no software I have ever come across like Emacs. It takes a lot of work to make it your own, and I totally get that is not for everyone. But if it is for you, everything else is too limiting.
(maybe to answer your actual question, I've never successfully made the switch. I always come back.)
Anothing thing from my recent (couple years) discoveries is that emacs is actually pretty good out of the box, so on my servers/containers I just started doing `sudo apt-get install --no-recommends emacs-nox` and use that. The only annoying default is the need to disable the "backup files".
I envy that with emacs you feel you only need 2 plugins. That probably operates with a ton more stability.
https://codeberg.org/dannyfritz/dotfiles/src/commit/b647b440...
Of course, there are other reasons to build your own configuration but I found purely from "IDE as means to an end" perspective getting into Neovim rabbit hole is a bit of waste of a time
When newer developers come to me looking for a recommendation of an editor to use, I don't often recommend Neovim, and I pretty much never recommend Emacs. I do recommend learning Vim motions, however, and my go-to is VS Code with the vscode-neovim plugin (uses actual nvim binary under the hood—no crappy emulation slowness/inaccuracies!)
For me personally, I've never met a situation where my Neovim or Emacs configuration couldn't handle a refactor or some other thing an IDE does. And then with all the other things like search/navigation (Consult+Orderless in Emacs is unparalleled) and working with files, git, or the filesystem (dired-mode in emacs, Oil.nvim for Neovim)... I'm just so much faster in one of those two editors than I am in any IDE. The ability to do everything you need in a single tool using the same paradigm and keybindings and concepts to get all of it done is pretty powerful.
I do also have a nervous aversion to using products that are not fully open source. Probably an OCD fear of losing full control over my editing and development experience.
reminds me of combobulate (which i haven't tried but think looks cool) but sounds much more light/limited
Apparently this was changed only recently. I am surprised not many people know/talk about this change. Still looking for a fix.
Edit: I do agree it is annoying but for unknown codebases it helped me a lot.
I also switched from lsp-mode to eglot and it's better. Less fragile.
I agree with this, and it's a big reason why I made the switch myself. I also find myself favoring lightweight (and especially builtin) capabilities, because really large, comprehensive plug-ins tend to have their own paradigms or ways of doing things that break from the super well-designed Emacs system of everything just working together magically.
One thing I've struggled with is simply remembering everything it can do. I made a Desktop Mat[0] to help remind me ( literally just finished it so will see how helpful it is when the printed version arrives ).
I'm still editing my config every now and then, but mostly smoothing out interactions with the software. Latest was configuring magit to have a speedier interaction by removing some stuff I don't care about.
When VSCode came about and had great integration for a language I was primarily using at the time ( Typescript ) I switched to that and was happy for the comparatively low config required. My Helix switch came about just because of how VSCode absolutely chewed through battery usage when I was out and about.
I tried helix last month and was immediately hit by how simple it is to get started. It still doesn't come naturally to me, but I managed to quickly get used to the basics: jumping around code, search, yanking/pasting stuff, and switching buffers and windows. I have no idea about its history, but Helix is a very well designed piece of software. Global search is particularly nice, lsp integration just works, it's also exceptionally fast. It feels very nice and comfortable when a software has sane, coherent, and actually-helpful defaults.
I'm definitley going to keep using it and getting myself more accustomed to this way of editing, I will keep emacs as my default, but helix is just so fast it might become my go-to over time for editing code.
- if a main point is modal editing, vim lets you use your muscle memory everywhere, since vim mode is pretty common (plus extensions like vimium, and almost certainly having vim on any machine you ssh to).
- helix seems to be both opinionated about simplicity (no terminal integration) and not very extensible with plugins (no linting unless it comes from the lsp)? That seems to be a limited combination, forcing you to have your whole setup built upside the editor (tmux etc).
I’m not saying this to disparage the project, just wondering: what are the big upsides that balance those cons?
There's not a lot of devs who want to directly use the basic vim experience (without any plugins/config). Even with similar keybinds, most devs aren't going to want to switch between editors. Instead, most devs are going to want to use an editor with a personalised setup and set of capabilities. The exception are those few devs who are actually sshing between many machines with a raw-dogged editor experience because it's a hard requirement, in which case they can continue to use nano, vim, vi, emacs, ed, etc...
> - helix seems to be both opinionated about simplicity (no terminal integration) and not very extensible with plugins (no linting unless it comes from the lsp)? That seems to be a limited combination, forcing you to have your whole setup built upside the editor (tmux etc).
Helix was actually built with the intention of having a whole setup built inside the editor, but that's not what the community wanted. So helix has been transitioning into adding a plugin system so the community can turn it into an all-in-one editor. The plugin system is not enabled in the main releases yet, but it's been in it's own branch for a while with a community of plugins already developed and with devs using them now. But all things considered it is the right move to hold back the release and get the shape of the plugin system as good as possible, because if it were released then needed to be changed, you'd be negatively affecting all the plugins that were released before the change.
---
The biggest upside is fixing the poor UX decision that's been baked into vi and carried into vim and neovim. That being, the order of operations which make it harder to see your changes before you commit them, making unintended side-effects easier to occur.
My current vim8 setup is very simple and has served me well for 8+ years. I'm currently using vim8 because it's what is available within my LTS distribution.
I only use one automatically loaded plugin (`vim-tmux-navigator` to simplify moving between vim splits and tmux splits, it's a passive plugin). I have reviewed the code for the plugin and I don't update it.
I only use two "optional" plugins (you can enable them using vim's built in package manager with :packadd!). I use `ale` (lsp, diagnostics and automatic formatting on save) and `vim-fugitive` (git workflow inside vim).
Why use ale? Because it works with vim8 and doesn't require any other dependencies like nodejs or python. I have reviewed the code and I'm happy with it. Install it and use it, don't update it without a reason.
Why vim-fugitive? It's a productivity booster. You install it once and forget about it. Tim Pope is the man.
I don't automatically load plugins (besides tmux-navigator) for a few reasons. Most of the time when I'm using vim it's for quick editing where I don't need the heavy tooling (like git integration or an lsp). If I am working on a project with a long term session I will enable git and lsp.
There's no need to automatically run code unless you need it.
I have come to love modern Neovim so much that I wrote a simple Python package[0][1] to solve this issue, at least for folks who already have to work within the Python ecosystem and use tools like `uv` or `pipx`.
You can install Neovim 0.11.4 (latest version as of building that package) with `uv tool install binwheels-neovim` or `pipx install binwheels-neovim`, and quickly be up and running with a modern Neovim on Windows, macOS, or Linux.
The package just wraps the official Neovim releases and relies on uv or pipx to select the right binaries for your operating system and architecture. The exception is Linux, where I actually build the binaries from source in a manylinux container[2] because I wanted older GLIBC support than the official Neovim releases are compiled for. Was tired of the official releases not supporting the older Linux systems I often need to work on.
[0] https://pypi.org/project/binwheels-neovim/
[1] https://github.com/wilkystyle/binwheels-neovim
[2] https://github.com/wilkystyle/binwheels-neovim/actions/runs/...
For people who use helix and want the full IDE experience, why not Zed, or maye even VSC/JetBraind IDE (come to think of it, how's fleet doing?).
If i need something simple I fall back to nvim, and if I'm missing features I sometimes boot up WebStorm (or if a colleague wants to navigate through something)
EDIT: for context I'm a hobbyist who use n/vim for 12 years or so before switching to helix for the last couple of years. There are several things about nvim behavior that I still miss quite a bit and still feel more natural, but the instant startup of helix vs a second or two for nvim configured to a similar level of functional makes it totally worthwhile for me.
Part of it for me is that I really don't like the massive plugin / config wall that is the modern neovim/spacemacs experience. I truly appreciate the extensibility and the fact that those plugins allow for things helix just can't do right now (e.g. emacs is effectively the IDE for any lisp language and you can't load up a repl in helix). But the problem with that becomes not only do I need to learn the application itself, but now I need to learn a whole bunch of plugins, and often learn about the differences between them because everyone has their own preferences in combinations, so looking up how to do something covered by a plugin often leads you to multiple incompatible answers. And even if you bite the bullet on trying to build your own config from scratch (which I did with emacs) you're still often just taking on faith other people's recommendations on stuff, or spending a lot of time learning about a plugin to make a decision and that time is time you're not spending on whatever you want to be doing.
Helix benefits from being new and not carrying generations of legacy decisions behind it. As a result, helix can make decisions and defaults that fit "modern" expectations. It's not perfect by any stretch of the imagination, but it's also probably what I would recommend for someone "new to the TUI" as a place to start when you just need something with a good chunk of modern conveniences and don't want to spend the next few weeks getting the editor configured.
It’s only a matter of time before someone deploys something malicious to one of those 50 plugins.
Language support in JetBrains IDEs is top-notch, but they consume a lot of resources and do not work well for remote development in my experience. I still open them from time to time to do complex refactoring. Also, I just hate the idea of opening Goland, RustRover, and PyCharm at the same time. I have a monorepo at work with multiple languages.
I used neovim for 20 years and still like it a lot. But after some plugins broke I wanted to give helix a try. I am missing a plugin system in helix.
However, for me, helix comes with the nearly perfect amount of functionality, while being extremely responsive. It also made me appreciate stuff like multicursor, which I haven’t tried before.
I tried multiple out of the box nvim solutions, but never liked one of those.
I looked into zed, but don’t see much of a reason to use it. Maybe I should give it a try.
Whoa neovim's been around that long already?
:reset-diff-change
git checkout -p
x Select current line, if already selected, extend to next line extend_line_below
X Extend selection to line bounds (line-wise selection)
Last I checked they still don't have a plugin system., I bet they'll see a jump in popularity when that gets going.
Neovim exist to improve upon concepts in Vim (easier plugin development with Lua).
Helix exists to improve upon concepts in Neovim (need for less plugins, more out of box functionality).
that apparently includes helix's creators, considering neovim is mentioned both on its website and in its repository's readme
Helix is more similar to Neovim than Emacs and Helix is newer and less popular than Neovim, so people are predictably curious on why they should choose Helix over Neovim.
I might try to learn this instead of neovim for a while.
Last time I checked Helix did not offer this. Has this changed?
:insert-output Run shell command, inserting output before each selection.
:append-output Run shell command, appending output after each selection.
:pipe, :| Pipe each selection to the shell command.
:pipe-to Pipe each selection to the shell command, ignoring output.
:run-shell-command, :sh, :! Run a shell command
Helix is built around showing you what will change before you commit the change which reduces the accidental change problem that vi, vim, neovim has.
In the case of Vis' regex based commands, helix will let you do a regex based find, remove certain matches, then commit a regex replace.
Or you could check out the branch where the plugin system and the community's plugins have been kicking around for a while now