Hacker News Reader

Comment by dleslie | Hacker News Reader

> I switched from Linux to Windows

This is where many people leave Emacs. It's not that there's better tools on Windows, it's that Emacs on windows is terrible.

Magit is orders of magnitude slower, for instance, and external tool integration is grossly impeded by the poor availability and quality of package management on Windows.

Maintaining Emacs on windows takes considerable effort.

Personally, I make do with WSL1 and Emacs. WSL2 is intolerably broken when working on NTFS folders, so it's a nonstarter. This means Docker is broken (mounts don't work) but Emacs works well.

Yah, it sucks.

View on HN

> it's that Emacs on windows is terrible.

I use Emacs on Linux and Windows equally. I've used Emacs on Windows at work for a decade. Other than magit, there's virtually no difference in experience between the two platforms.[1] Of course, perhaps you use a different set of features than I do.

> Maintaining Emacs on windows takes considerable effort.

I install it on my work laptop every 2-3 years, and for the last 5 or so years, it's literally been "Download, unzip and run." Occasionally I have to make registry changes to get org-protocol to work. Even 10 years ago, it was still fairly simple - just a bit of extra work to get some GTK related stuff installed as well. But right now it couldn't be easier.[2]

And I don't even use WSL. This is native Windows.

[1] Of course, if your Emacs is calling out to tools like grep/find, etc, you'll have issues unless you install those tools. Although these days, it's not hard to configure it to use ripgrep and fd, which has excellent Windows support.

[2] And perhaps installing LaTeX if you want to export org files to PDF.

Mostly issues with lsp servers; particularly ccls and omnisharp.

For whatever it's worth, I've found the same thing: that the amount of effort required to use Emacs on Windows is minimal. Since Windows is non-POSIX, inevitably some things are different, but it doesn't take much to deal with this.

My .emacs.d folder (now 15 years old, and a bit of a monster, a whole git repo with thousands of lines of nonsense plus a pile of submodules) has no more than a few spot hacks for Windows vs POSIX vs macOS.

(The most annoying bit has probably been keeping on top of the font names! - but interestingly these are actually super easy to deal with on Windows. It's Unix that's the pain in the arse here.)

I've been using the mingw windows version every day for years. I used to spend time now and then to make new things work, recently I switched to doom emacs, and almost every new functionality or package I try works out of the box.

It's true that Magit is slow, but I still prefer it to the command line.

I'm going to give Doom a shot; I ignored Spacemacs because rebinding native keys and forcing evil-mode was a nonstarter for me. If I wanted to use Vim then I would, and did for a time.

... and that was an immediately disappointing experience.

1. Saving customizations didn't work.

2. Launching Emacs as a client that auto-launches a server _always_ failed the first time.

3. I really don't need a custom package manager built on top of another custom package manager to replace the core package manager. straight.el would've been enough, thanks.

4. My batteries-included Emacs launches a server in a few seconds, a client almost instantaneously; doom took minutes to launch a server on my Pinebook Pro and clients took a few seconds.

5. Another needless dotfile dir in my home. ~/.doom.d could easily have been ~/.emacs.d/doom.d

"forcing evil-mode"?! Spacemacs asks if you want to use "evil", "holy", or "hybrid" mode early in the process.

Interesting, I suppose I missed that when skimming the GitHub repo readme.

Well with wsl2 I gave windows a go for my work dev environment. And so far emacs had been amazing via wsl2 and a x server.

I should note I did modify the xserver to allow me to toggle individual windows going fullsceen. And am running native emacs (which is blazing fast)

WSL2 + X server was my go-to for a few weeks, but then a rather large repo I access, with git-lfs, became corrupted due to WSL2 + NTFS bugs. They've been mentioned on HN before.

Thereafter, every day or two the same bug would arise and I'd have to lose an hour or two recovering the repo. Just a week or so ago I switched back to WSL1, and I ssh into an RPi4 to access docker containers, using qemu and binfmt to execute x86 code slowly.

Yah, it sucks. ;)

Reply on news.ycombinator.com