What features and expectations do you have for Neomacs?

[eval-exec]

Hi all! 👋

I’m developing Neomacs and want to hear from you:

• What do you use Emacs for?
• What features or improvements do you expect from Neomacs?
• What matters most to you in a modern Emacs-like editor?

Please share your thoughts — even brief replies are super helpful! 🙌

[galoistom]

I mainly use it for note taking with org-mode (i dislike obsidian as it is not open source👎, and i find it much easier to write plugins for emacs), and occasionally write some code and edit configs

I hope the image support to be better and configuration to be easier, i'm still unable to manually support tikzcd image rendering😭 (i use math-preview to get a faster inline preview experience, and it use mathjax as backend, so i have to handle tizk manually), and perhaps allow the orgmode to have a more intuitive view, let centered block to be centered for example. A better browser, and pdf viewer is also derirable, as i some time have half the screen showing text book/webpage and note on the other (eww is dreadful at viewing sicp QWQ) .

[eval-exec]

I hope neomacs can support all of these needs too. 💙

[junghan0611]

From an Emacs core researcher: respect for the characterization-testing approach

Hi @eval-exec,

I'm an Emacs user and researcher from Korea, working on a related project
(incremental Emacs modernization via Zig — different path, same mountain).

I studied the Neomacs codebase in depth — not just the Rust code,
but the infrastructure behind it. What I found genuinely impressed me.

What I noticed

The test/neovm/vm-compat/ oracle harness is, in my view, the most important
artifact in this repository — more important than the 261K lines of Rust.

The structure:

• Oracle: GNU Emacs --batch -Q as the truth source
  (oracle_eval.el, 34 lines — that's all it takes)
• Execution queue: docs/plan.md driving systematic progress
• Regression gate: make check-all-neovm-strict after every change

This is textbook characterization testing (Michael Feathers) with a live oracle,
and it enables a development velocity I've never seen in any Emacs-related project.
744 case files covering thousands of individual assertions, TSV-based behavioral
comparison, automated build-test cycle — this is engineering at a high level.

Why this is different from Remacs

Remacs attempted an incremental, function-by-function port from C to Rust,
preserving the original architecture. It stalled at the core (GC, bytecode).

Neomacs uses behavioral oracle comparison — the question isn't
"does this Rust code structurally match the C code?" but
"does this Rust code produce the same result as GNU Emacs?"

That's a fundamentally stronger guarantee, and it's what makes
the aggressive development pace sustainable.

My work and potential convergence

I'm working on i-am-emacs — a narrower approach: Zig-based I/O thread
separation and process-level concurrency (multi-daemon architecture).
Not a rewrite, but surgical patches that could merge upstream.

I see our projects as complementary:

• Your oracle harness could validate my Zig patches
  ("do all 744 case files still pass after I/O thread changes?")
• My research on feature/igc (MPS-based incremental GC, merging to master)
  may be relevant to Neomacs GC design
• If both succeed: GPU display (Neomacs) + async I/O (i-am-emacs)
  • incremental GC (igc) = Emacs reborn

What I'd like to learn

1. How you structure the plan.md execution queue for long autonomous sessions
2. Which Elisp subsystems were hardest to achieve oracle parity
3. Whether you've considered a shared oracle corpus that other Emacs
   reimplementations (Rune, etc.) could also use as a compatibility benchmark

Thank you for this work. If Neomacs succeeds,
it lights up 45 years of Emacs history.

— junghan (@junghanacs)
https://notes.junghanacs.com

[eval-exec]

Thank you. I think zig is a wise choice for implement emacs related project. Zig is easy to integrate with C.

How you structure the plan.md execution queue for long autonomous sessions

The plan.md is only work for elisp-vm compatibility test. I just to want to make sure neomacs is 100% compatible with official emacs.

Which Elisp subsystems were hardest to achieve oracle parity

interactive spec parsing, I think

Whether you've considered a shared oracle corpus

Yes, create shared oracle corpus would be very valuable.

[xaiki]

Congratulation on your effort, brilliant architecture and insane dedication.

When the nights get cold, i dream of:

• better font-face, or at least more compatible with the outside world, maybe things like svg-tag-mode can be supported as part of the core (so they are faster too).
• native crdt multi-host sessions: you can build something like that with https://github.com/zaeph/crdt.el and tox or tailscale, but i'd love it to be a core part of emacs where i could just share all my buffers with an optimized protocol over many machines (connects to the next point).
• mobile apps (long shot), i'd love to be able to make little org-mode apps for my phone that could just bundle your rust core and easily have me write elisp as a first class mobile citizen.
• rusty icons, here a proposal:

[eval-exec]

Thanks a lot. This is super helpful. Better font/SVG compatibility is on our near-term list; CRDT/mobile are longer-term, and your icon concepts look great.
Also, neomacs using winit+wgpu, and we can compile neomacs to mobile device, and wgpu support WebAssembly, so I think neomacs can also be run in the Web Browser.