mintflow

250k for openwrt based risc v router? Maybe need do more work such as using vyos + fdio/vpp

This is cool, let aside the token usage, perhaps it can help analyze tcp throughput by redirect wire shark/to dump result

Finally a tailscale rust port is coming, i think it's will make build app with builtin tailscale connectivity more easily compared to libtailscale

As a developer who have been built some tailscale-based clients, I think this maybe acceptable because they running a business with money from the VCs.

And I am also very grateful that tailscale implement some workaround for systems such as apple-based OS with core APIs built into the open source code, thus if you really need you can just look the open source code and doing accordingly, though it really need some research work.

For the long term if they really do not want to open source the core client code (which I do not believe at the moment), I think support a fully open source coordinator and open source client based on the fork will still be doable.

Have built a cross alternative tailscale gui client based on tauri, the rust and ffi to cgo tailscale feel a little tough, I was wondering it will save a lot time to me if the tauri had been written in go.

Seems Miguel’s velox point a new idea, leveraging the wry and use ffi to go, and rewrite some tooling.

I hope I will have the spare time and energy to give a try…

just did some spec reading, it's quite clear and nit.

I can understand that put UDP payload into a single HTTP stream, at least when QUIC transport is in use, there is no UDP in TCP case.

The Source Address/Port in the UDP payload message serve as key to handle to the tunnel client if I understand correctly?

It’s great for you to open source the protocol and implementation, it written in rust which I will definitely consider to learn it add add to my vpn client in the future

After reading the article, I am wondering is that is there no test case to coverage the behavior that modify the CNAME order in the response? I think it should be simple to run a fleet of various OS/DNS client combinations to test the behavior.

And I also being shocked that Cisco Switch goes to reboot loop with this DNS order issue.

I like this idea but really do not want to share my personal data to cloud based LLM vendors.

I have a folder which is controlled by Git, the folder contains various markdown files as my personal knowledge base and work planning files (It's a long story that I have gradually migrate from EverNote->OneNote->Obsidian->plain markdown files + Git), last time I tried to wire a Local LLM API(using LMStudio) to claude code/open code, and use the agent to analyze some documents, but the result is not quite good, either can't find the files or answer quality is bad.

This is the reason why I am creating a Debian VM on my macOS to let Claude code in yolo mode to do some experiment:)

Tailscale is quite handy in remote agent coding, Sometimes I use tailscale and RustDesk on my phone to check Claude code, I also built an app called NovaAccess which bake tailscale into the app which does not confict of VPN I used.

Apple still get quite good revenue and best in class engineers, how does this allowed to happen?

I have a software that need to build aarch64 (for some aarch64 box with 4 core cpu), currently using Oracle cloud's 4core24G Arm neoverse n1 as github self host runner to build it.

Seems this machine is more powerful than it, definitely attractive to me for a physical aarch64 self host runner.

great project, just add the mcp and try in claude code, it automatically help me to broser a local forum and give a summary of hot posts today.

this is the second MCP i added, quit impressive.

Merry christmas!

Just noticed that color of the HN top bar is more red which I never noticed in previous years.

Merry Christmas to everyone.

recently I just noticed github trying(but failed) to charge the self host runners, I find a afternoon to setup a mini PC to install freeBSD and gitaea on it, then setup tailscale to let it only listen on the 100.64.x.x IP address.

Since I do not make this node public accessable, so no worry for AI web crawlers:)

amnezia-wg is quite cool and they have built the kmod too, I did some test so far they can works even in my location which block wireguard server quickly.

I was thinking that's hard, but I noticed that vpp get ported to FreeBSD using epoll shim library, and I learnt apple Darwin use some some userland of FreeBSD to do POSIX compatibility, then after some tests and hacking, most related to minor POSIX API adaptation such as mmap and one major coroutine need add some assembly code, and it work! But I think most disappointed to me is that apple do lack some vectorized network IO unless do some kernel extension or other sort non standard ways.

For the similar reason I do not using any go based proxy code in my MintFlow app, and use rust to implement some proxy protocols.

But my app’s wireguard is natively implemented by fdio vpp plugin, so it’s based on C.

This sucks, it make me feel so silly after decide to move back to github self hosted runners just because I do not want to run act on a remote ARM64 server.

I was just using act (https://github.com/nektos/act) on my local server to build the X64 packages for my project, since I want to streamline it with ARM64 support, I migrated to the github self hosted runners.

This is really ridiculous, is M$ really lack that money just to schedule the Jobs running not in there infra?

Working on my app https://apps.apple.com/us/app/mintflow-netstack/id6742394218

As an engineer working on networking and fiddle with various networking OS on router and switch, I finally port my favorite fd.io vpp to darwin platform and built a app to management multiple VPN/Proxy in one profile.

Also in this project I start writing some rust code with many years experience in C but rust's memory and high performance really impressed me a lot.

Great project, I really love tree-sitter, recently I added a ini variant config profile support to my app, and just use gemini to write a grammar and combine it with another great project called runestone to support highlight the config profile, the total progress is quite smooth.

as a tailscale user and quite obsessed with tailscale related news, it's quite funny to see the tailscale on the starlink dish news...

Recently my multiple VPN server nodes just randomly cannot connect to cloudflare CDN IPs, from different provider VPS, while the Host Linux network does not have the issue; vpp share the same address with Linux and use tc stateless NAT to do the trick.

I finally work around this by change the tcp options sent by vpp tcp stack.

But the whole thing made me worry there must be something deployed which cause this issue.

But I do not think that related with this network issue, it just reminds me the above, I feel there are frequently new articles about cloudflare networking, maybe new method or new deployment sort of related high probability of issues

Exit node really is a handy solution for build private vpn for sharing.

I have build a vpn called Echo VPN for apple platform which actually use tailscale open source core.

Also I think there is another benefit is that wireguard can be DPIed easily now adays, but DERP leverage HTTPS and upgrade which can do some obfuscation too

I think you need open UDP 12345 to public internet, where the Tailscale Disco protocol runs on that, which is the underlay port.

When I was port fd.io vpp to apple platform for my App, there is code that's implement coroutine in inline ASM in a C file but not in Apple supported syntax, I have succesfully use Claude web interface to get the job done (Claude code was not yet released), though as like in this article, I have strong domain specific knowledge to provide a relevant prompt to the code.

Nowadays I heavily rely Claude Code to write code, I start a task by creating a design, then I write a bunch of prompt which cover the design details and detail requirements and interaction/interface with other compoments. So far so good, it boost the productivity much.

But I am really worrying or still not be able to believe this is the new norm of coding.

I was also used evernote, then onenote, then notion and obsidian to track my TODO items and personal knowledge base, finally settled using pure local markdown file and edit it using emacs, for syncing i just init a git repo and sync to private github repo, so far so good.

Maybe as a coder, I will someday will just calm down and write code without using any LLMs, just for train my thinking capability...

just curios, will it be supported on iOS, it would be great to build local llm app with this project.