1. Sockets and async (start here)

• std::net (TcpListener, TcpStream) for blocking basics.
• Tokio for real work. Its official tutorial (tokio.rs/tokio/tutorial) builds a mini-Redis server and client, which covers framing, shared state, and graceful shutdown.
• tokio-util codecs (Framed, LengthDelimitedCodec) for protocol framing, and the bytes crate for buffers.
• socket2 when you need options std doesn’t expose (SO_REUSEPORT, keepalive tuning, buffer sizes, TCP_NODELAY variants). mio if you want the epoll layer under Tokio.

2. Packets and raw access

Assuming you mean Rust-for-Linux (in-tree kernel modules), the usual workflow is a dev shell for the build plus QEMU for testing, rather than booting the kernel on your NixOS machine.

1. Dev shell (shell.nix, channels-style):

{ pkgs ? import <nixpkgs> {} }:
pkgs.mkShell {
  nativeBuildInputs = with pkgs; [
    gnumake flex bison bc pkg-config ncurses
    llvmPackages.clang llvmPackages.lld llvmPackages.libclang
    rustc cargo rust-bindgen rustfmt clippy rust-analyzer
    openssl elfutils perl qemu busybox
  ];
  RUST_LIB_SRC = "${pkgs.rustPlatform.rustLibSrc}";
  LIBCLANG_PATH = "${pkgs.llvmPackages.libclang.lib}/lib";
  hardeningDisable = [ "all" ];
}

hardeningDisable matters because Nix’s default hardening flags break kernel builds.

Finally I have set up my markdown properly.

It starts with Helix: Setup for Markdown. It involves installing dprint. Although it mentions only helix, I have created similar configurations for zed as well.

Below are the configurations files for the respective editors.

languages.toml for helix:

[[language]]
name = "go"
auto-format = true
formatter = { command = "goimports" }

[[language]]
name = "rust"
auto-format = true

[[language]]
name = "c"
auto-format = true

[[language]]
name = "markdown"
auto-format = true
formatter = {command = "dprint", args = ["fmt", "--stdin", "md"]}

settings.json for zed:

In the last two posts, I wrote extensively about emacs. However, today I have come across this post. I have learnt the term .emacs Bankruptcy.

In his case, over the days, the configuration of his emacs is getting bigger and bigger that needs a reboot, or fall under its own weight. He also doubts that whether he needs the power of org-mode. He repeatedly emphasizes that he likes emacs. At some point during last week, I even seriously thought about getting a 10-inch chuwi just for emacs. My goals was to have a dedicated emacs “appliance” so that I can carry it anywhere. But nothing beats the wake-up speed, or snappiness, of a macbook. So I gave it up quickly.

I am writing in emacs org mode again which I quite enjoy doing.

In the last couple of days since CNY, my focus is emacs. I RTFM about it, try to use it and even compile from the git repo.

I recalled my memories about the emacs built-in tutorial to gain basic keyboard nevigation and editing. Then read about the manual. It is a quite an undertaking. I just glanced through it and get most of my basic editing tasks done. I am sure that I will come back to it on and off.

I am writing something again. The last two years have been rough for me. Ups and downs, changes in the family etc. Now I have got enough time to write about something.

I would like to write about emacs. I am writing this post in vanilla emacs configuration. Even spell-check is not working properly. I am doing this in a terminal, on a mac. There are a lots of alternatives such as byword. Why I am torturing myself with a 40-year old text based software?

Finally I could write something again. I am writing this in markdown, my editor is vscode and my keyboard is nuphy 75 old edition.

I have been using the nuphy 75 old edition for a while. Then I have facinated by 60 keyboard. I have tired the nuphy 60 old edition. My only complain is that both the ` and the ~ characters are supported by the escape key. It is a bit annoying when typing ~ for home directories. So I go for the 75 keyboard.

I am a bit excited.

I have been wandering and straying during my way of learning C++. On and off I feel at a lost somehow. I have tried to explore different use cases and code bases of C++ projects. 10 C++ open source projects welcoming contributions says ClickHouse is one of them. It is a database project that unlike any other relational databases that I have learnt back to school. Learning C++ and another database paradigm, sounds cool.

I have tried my best to complete the My First Language Frontend with LLVM tutorial as a learning path to llvm. It is quite an undertaking. I have nevertheless completed it, but without completely knowing what I am doing. So it has taken longer time that I think. I have also picked up other hobbies, so the updates has not be frequent.

It involves more things that I do not know. Other than C++, the author of the tutorial assumes the reader has some basic compiler design knowledge which I am lacking. The code snippets presented in the text are different from the complete code listing at the end. It seems that the content of each chapter is built upon the previous one, obviously except chapter one. Most of the time it is, but not at chapter 9. For chapter 8, I could not even compile the code listing. I changed the file name to t.cpp. It says