Hi everyone, as a software engineer, I wanted to know more about networking and BGP. So I started implementing BGP about two years ago. But it stalled at a message parser, because I had too much knowledge gap. I stopped. But late last year, I started again, this time with AI. I studied up the RFCs, designed the architecture and Claude did most of typing. I read every RFC and every code, which is where learning happened.
7 months later, it's live and I'm running AS4242423930 on dn42, carrying the full routing table (~2k prefixes) with two eBGP peers. I'm also running two "POPs" in US east and west coast, running iBGP. It doesn't actually program routes to kernel yet (it's just a speaker). I figured it's enough for route servers or software controllers that don't actually forward packets.
As for the project, what I wanted to focus on was observability. This natively emits operational metrics as CloudWatch EMF (for AWS), JSON (for other Cloud), and also Prometheus. Memory-safety and efficiency were also the goal (Rust). And the project is actually not just BGP, but a routing stack. Because I'm also curious about protocols other than BGP.. So I'll implement more routing protocols as I continue learning more. I named it Rogg ("Routing, gg") which consists of bgpgg ("BGP, gg") and ggsh ("gg shell"). I grew up a gamer and couldn't resist. Thanks for checking it out.
Hi everyone, I wanted to share a language I made that only uses a series of bracket characters. All operations are prefix-free, meaning one can follow another without ambiguity.
Same. Having finished the course years ago, I remember most vividly the following: never implement your own crypto. Not many details survived the time but it was intellectually rewarding.
Interesting point. While I sympathize with the cause, I personally wouldn't find it the best money spent, had I donated, because it is impossible to measure the impact of the money spent.
Thanks for the compliment. It is possible to use any other methods to synchronize notes because Dnote stores everything in a single SQLite file. The dedicated server becomes useful for providing a web interface and API access to the notes.
Adding lines such as "topic:..." sounds like a very useful and low key solution. I might start trying that. Thanks.
You can use Dnote web app on the phone to do full text search and list notes by book. You can also install the website as a progressive web app for usability.
On the downside, using this web interface subjects users to a third party system (albeit open source) and forces them to give up the autonomy that they enjoy by simply using plaintext files.
> I tag each text file with topic, date and context keywords.
Nice idea. How are you doing that? Also, do you version control your notes?
> What does not work great is on the phone
I maintain an open source project called Dnote (https://github.com/dnote/dnote) which solves this problem for you. It's basically a command line notebook using SQLite + a mobile friendly web interface to which you can sync your notes.
Agree with you that we should avoid being locked into a proprietary formats or platforms. Businesses and platforms come and go, but our notes should stay as readily accessible as possible.
Very nice. What I liked about it is that thumbnails to the article allows readers to skim through the list more quickly and still understand more clearly what they are about.
Personally I have noticed some changes, but nothing that impacted my productivity in any significant way, and paid no more attention.
I believe sometimes we tend to overreact to certain changes that have minimal impact on our lives, because of our attachments to the tools. For instance, on HN, we seem to get a sea of "That's it I'm moving to Firefox/GitLab/etc.." comments often when their counterparts change something. Sometimes those reactions seem warranted, and in this case, not really.
Beanstalk is cool. Just curious, in what ways could it be more simplified? I think we can already just upload our code and forget about it. It could be that I am missing something.
Good to know that the owners are "pretty cool and nice" but that is orthogonal to the point.
> They also have an option to destroy all the data they have on you.
Unless open source, the users never know if the data is destroyed or not. Furthermore, it is impossible for the users to verify the checksum of software running on the server side. Therefore on-premise install could be a nice solution.
Looks like a cool service but sounds somewhat off-putting in that they will have all those data about my behavior. Maybe it would be cool to be able to self-host.
Yeah, Hub is such a valuable tool for automating common tasks on GitHub such a releasing. Hopefully GitHub CLI will maintain a feature parity with Hub so that people can transition smoothly if needed.
7 months later, it's live and I'm running AS4242423930 on dn42, carrying the full routing table (~2k prefixes) with two eBGP peers. I'm also running two "POPs" in US east and west coast, running iBGP. It doesn't actually program routes to kernel yet (it's just a speaker). I figured it's enough for route servers or software controllers that don't actually forward packets.
As for the project, what I wanted to focus on was observability. This natively emits operational metrics as CloudWatch EMF (for AWS), JSON (for other Cloud), and also Prometheus. Memory-safety and efficiency were also the goal (Rust). And the project is actually not just BGP, but a routing stack. Because I'm also curious about protocols other than BGP.. So I'll implement more routing protocols as I continue learning more. I named it Rogg ("Routing, gg") which consists of bgpgg ("BGP, gg") and ggsh ("gg shell"). I grew up a gamer and couldn't resist. Thanks for checking it out.