All Beej's guides are fantastic, but if you're interested in Network programming you could do way worse than follow this up with his guide to network concepts. https://beej.us/guide/bgnet0/
If you search Algolia for Beej you'll see his material has been on hacker news numerous times.
Anyone knows of a well-rounded guide that explains the structure of the Internet at large? I know everything listed in the one you linked, but I have a nebulous idea of how anything works after my packet exits my gateway.
So, BGP, autonomous systems, peering agreements and other site-to-site routing protocols that keep the entire Internet infrastructure ticking along.
I don't know of any one resource that gives a overview of everything, I find that browsing Wikipedia for each topic gives a good overview of each topic though, with the overview being the "Internet Architecture" category: https://en.wikipedia.org/wiki/Category:Internet_architecture
Some bookmarks I have regarding BGP and peering that dives into it a bit:
Be suspicious of anything that includes DNSSEC in the "tech fundamentals", since it's rarely implemented (low single digit percentage uptake in the major TLDs). What other complexity are they failing to screen out?
Agreed and yeah he mentions that. I think he mostly just makes you familiar with DNSSEC because he sells AWS certification courses and it will come up on the Networking Specialty. The entire segment is only like 90 minutes so it’s not crazy in depth.
Or, as I just recently learned, you can click on the (beej.us) domain name next to the title at the top of this page and it will take you to a list of all the HN submissions for the site:
I agree with everything you said. Additionally, I think that the ZeroMQ guide is also a great resource to learn a lot of the common networking patterns, even if you don't plan on using ZeroMQ directly.
There is also a C guide [1] which has a pretty interesting email policy...
I’m generally available to help out with email questions so feel free to write in, but I can’t guarantee a response.
I lead a pretty busy life and there are times when I just can’t answer a question you have.
When that’s the case, I usually just delete the message. It’s nothing personal;
I just won’t ever have the time to give the detailed answer you require. [...]
I had a work colleague with a similarly refreshing Out-Of-Office E-mail autoresponder that included something like: "I'm on vacation. When I get back I'm not going to have time to read all my missed E-mails. So your mail is going to be ignored and deleted with the other 2,000 unread mails. If it's important, E-mail me again after I return on [date]."
Here's a product (Gmail/Outlook add-on or so): have your own LLM trained over your public texts (articles, books, blog entries, READMEs, Hacker News comments, and so on), mark the emails you can't respond to with the "GPT" label which puts the mail in a queue and if you don't remove the label in 7 days or so, it sends a response with a polite intro "I didn't have the time to get to your email, but here's what I might think as a response".
Good thing that with the coming devastating job purges due to statistical learning such "careers" based around bullshit jobs [1] where some words in a generated email can be "career-terminating" will be forever gone.
I think I'd argue that the people whose emails can be sufficiently answered with an LLM are the ones whose jobs are first on the chopping block, so the product has a self-destructing user base.
Not necessarily. When I was writing the initial comment I wasn't even thinking about job-related mails, but at someone like Sean Carroll who does a regular Ask me Anything [1] where lots of questions are skipped for lack of time, or in general academics who receive plenty of emails about articles and various field-related questions.
I can speak from experience that posting that type of material will generate some interesting email inquiries. I used to get email messages with a 4000-line program, multiple datasets, and "My program doesn't work. Can you tell me what the error is?" The awkwardness of deleting the message is less than the awkwardness of refusing what someone else views as a 30-second request.
When you know you definitly do not have time for an answer deleting it might unload your mental burden. Keeping hundreds of unamswered messages around might feel like an unresolved issue.
There's no functional difference between "not looking at it ever again" and "deleting it".
You should understand when they say "delete" that they mean "remove from their queue", whether it still exists in an "archive" folder, backups, trash, ... is irrelevant.
> "The way I see it, it's the educators who hold all the power in the world. No, stop laughing and hear me out. A good instructor can, over the course of their career, create hundreds of millions or even billions of dollars of GDP. Let's say a teacher teaches 40 students a year for 30 years. And Each of those students goes on to work for 40 years. And, due to the quality of instruction, they earn $10,000 more per year than they would have otherwise. That's $480 million in extra revenue. From one good instructor."
OK but providing value isn't the same as having power. It's almost the opposite thing.
When people talk about having or not having power, they rarely mean magnitude alone. The more important component is control. Can you wield your power to achieve some specific goal of yours? Can you use this power to overall improve your life, and lives of those you care about? Can you use this power to protect yourself from someone else using their power to achieve something at your expense?
Beej's teaching power only answers "Yes" to the first question here - and that's only if your goals have something to do with teaching or improving the world overall, somewhere, for some people.
I'm not sure I understand your third question, but I think beej's teaching power fulfills the second question in a very radical way which is that it is an act of caring for a large audience of strangers and does improve their lives.
Power that isn't attached to the desire to reap a direct benefit is a strange thing to witness, but it's power nonetheless!
It's whether or not a particular type of power you have can help you and yours in adversarial situations.
Can your Power of Teaching help you if a local thug wants you gone? Or, can your Power of Teaching help you when local governor decides to close down and raze your school, to be replaced with a factory? In both cases, someone else is exercising their power to directly or indirectly take away your livelihood (or life) - can your power prevent it, or at least mitigate the blow?
No power is good against all other powers, but some kinds of power are better at this in general. For instance, a wealthy businessman or politician could answer "Yes" to both scenarios above (replacing "school" with closest equivalent) - their money and influence can buy protection and work as a deterrent against the thug, and it can outweigh the influence of the local governor (or just allow for an effective bribe).
> I think beej's teaching power fulfills the second question in a very radical way which is that it is an act of caring for a large audience of strangers and does improve their lives.
Caring about large audience of strangers (or humanity in general) resonates with me, so I'll give this a passing mark here :).
> Power that isn't attached to the desire to reap a direct benefit is a strange thing to witness, but it's power nonetheless!
True. What I am trying to highlight here is that, when people talk about "power", in the sense of seeking more of it, or comparing people by how much of it they have, they usually don't count magnitudes of raw power, but rather how generic and how controllable it is.
This is similar to wealth, btw. Having $1M in cash makes you more wealthy in the immediate term than someone who has $1M in stocks, equipment, etc., as that other person won't get anything near a $1M if they try to convert their assets to cash.
Yea he's using economic output as a measure.. then equating it to having power, which is laughable.
Anyway, in an age of information abundance and videos, educators are overvalued if they just give instruction alone. It's the whole package, the checks and balances on the student, that possibly provides some sort of value. Then again, there are irregularities, like the programmer who's never been to college who contributes as much or even more.
Posted a million times before, and should be posted a million times again in the future. This is one of the only reference books I've ever used, and it got me through my first real programming job. Wonderful text. Thank you, Beej!
Open question: in what fields are you working where you really need to work with actual sockets? Most of the network programming I do fall into one of two categories:
1. A library exists to abstract over the bytes-into-sockets layer, meaning I don't need to deal with it; or
2. A library probably exists to abstract over the bytes-into-sockets layer, but I've chosen to ignore it either because I'm truly ignorant of it, or for intellectual stimulation.
(Note that in category 2 I count things like "not using protobuf/thrift/etc. for IPC.)
When are these things not true? Genuinely curious! I'm guessing interfacing with existing obscure third parties, or using obscure language environments, but what are examples of that? And what other usages are there?
I don't work with it either. But I find that learning the abstraction below where you're working at can be quite beneficial to understanding the constraints of your layer, debug, and solve problems.
E.g.: Learn the basics Transport Layer protocols (TCP/UDP) if you work with HTTP
2. Speaking unusual protocols for which (good) libraries don’t exist.
3. I think interacting with the sockets api can also be relevant for tweaking various options to get better performance (though often there are Linux defaults that can be tweaked instead)
at knownow i wrote an event-driven nonblocking http server in python with asyncore which could efficiently do comet over hundreds of concurrent http connections; existing http libraries were all blocking at the time
i also wrote a similar event-driven nonblocking http client in c for benchmarking and load-testing both that server and a different compatible one being written by another team in c++. this was nonblocking because it was important for the client to be faster than the server. at the time there was no libevent/libev/libuv so i wrote my own
at satellogic i maintained the upper layers of our cubesat space protocol stack, which provided a socket-like interface to our coap stack, which i also maintained part of
i also had to tweak kernel socket buffering settings for rabbitmq there
for wercam i'm experimenting with different approaches to ipc for efficiency; the data being transferred are pixel buffers so i don't think protobuf/thrift/etc. is going to help
for the mail server i share with some friends i wrote a milter in python to smtp-reject outside email to certain recipients unless it matched a whitelist; milter doesn't speak pb either and i couldn't find a python library
i wrote a multicast file transfer program in python (bccpo/bccpi) for copying files across the lan without having to fiddle around with ip addresses; the data being transferred is just (multicast) the ip and port and (unicast) the filename, file size, and file contents, so again i'm not sure pb would simplify anything. the hard part of getting it working was just multicast
i wrote a chat client for icb in python with asyncore, and again i couldn't find a suitable library
even when i'm using a library, it's often pretty useful to be able to read strace output or use select or poll with the file descriptor it's trying to hide from me
i don't think any of these qualify as either your category 1 or 2
i've also done a huge number of things that fall into category 2, of course, like writing an irc client in bash, writing a web server in assembly, and writing a chat server in three 80-column lines of c, writing a mud, and so on
I often work in a resource constrained environment where my program size is measured in two-digit kilobytes, RAM is measured in the single-digit kilobytes, and it can’t exceed some current measured in mA/h. When I am simply sending a UDP packet, I didn’t need nor can I afford the footprint that these libraries have.
It's useful to understand the socket API so that you can do things like read the output from strace, notice that accept() returns EAGAIN too often, and know what that indicates.
At least on Unix and Windows file read and write methods work for sockets too, so streaming bytes isn't different.
How would you implement, say, netcat? Launch a jango rest service with kubernetes, which will run in a wm if it's a wrong OS, create a certificate with letsencrypt and access it with a react electron app?
Why would I implement netcat when netcat is already implemented? That's sort of why I'm asking – at least in the circles I move, all the low hanging fruit are already picked, and picking the high fruit seems rare, so I'm interested in which people do and what those fruits are!
Had plenty of use cases where even zmq was overkill.
E.g. interfacing a research machine with an OEM machine or a machine to a standalone embedded target, sometimes it makes sense to just serialize, send over a socket, ethernet crossover cable and be done.
Your description makes it sound sort of like temporary code to perform a one-time (or few-times) job, and not something that's running and maintained for years. Is that a fair characterisation or am I misunderstanding?
I implement protocols to communicate with various hardware (RFID, LiDAR, etc). Most of these communicate over sockets. Some still use serial. I have a library I wrote of socket routines that I use, so it's not like I re-implement that every time, though. And for this, I'd really rather use my library than a 3rd party in case I have to dig in and see what's going on.
I've been working on the Protohackers [1] exercises in Elixir, putting my own spin on the problems by creating fault tolerant, scalable, high performance solutions-- and have been touching a lot of Elixir that I haven't had the opportunity to otherwise. The "Unusual Database" problem, particularly, has become a very deep rabbit hole assignment as it involves a UDP-based KV database. I stopped just short of clustering.
Writing servers in Elixir is straightforward because of the foundation exposed by Erlang's gen_tcp and gen_udp libraries. When you want to take your TCP server up a notch, try TCP acceptor pooling. There are two viable TCP acceptor pool libraries: Thousand Island, written in pure Elixir, and Ranch, written in Erlang.
Because of these well- abstracted foundation libraries and acceptor pool libraries, it's possible to write robust servers without venturing much into low level network programming. Beej explains the lowest level details well. I've done some of the equivalent work in Rust but found the slow progress of my work unsatisfying and moved back to Elixir quickly.
I was disappointed last night when I looked at the Protohackers website and it seems to be on hiatus. Such a great idea, but I imagine it's not trivial to manage.
There is a handful of people whose tabs I would instantly pick up were I to meet them somewhere because of the immense value I have gotten from their free work online, writing stuff, answering forum posts, etc.
And I couldn't be happier, Beej's guide is a seminal read for anyone looking to get into net programming. Exposing as many newbies to it as possible helps strengthen the field.
I'm not saying Rust is the answer to all things, I'm saying I want a Rust translation for continuity of the wisdom and concision I loved in this guide as a 14 year old.
The only problem is that it wouldn't be quite as useful. Zig, Rust, D, etc offer libraries and methodologies (async/await, for instance) that deprecate traditional raw socket programming. Instead, you would use something like tokio in Rust to do the same thing.
You certainly can program raw sockets, the techniques are almost orthogonal to the beej guide. It's the same API and Structs (using FFI) and you would change some glue logic to make it more rust idiomatic. It would just be less useful in those languages, generally.
I followed Beej's linked guide to transfer data over udp sometime ago, written in C on a Windows machine. I then compiled the code on Mac os x but could never get it to work properly. Tried all the Google suggestions at the time, posting this comment in case there's an obvious "oh did you set X?" tip from someone that has done a similar task.
Maybe the reason you're not getting help is is that you're not asking a question. You're not even describing a problem. You're only stating that you have a problem.
I love this guide, I gave this link as a reference to my students when I was teaching networks a few years ago. But the last time I looked at it the guide didn't have IPv6 taken into account! Very nice to know that it is now the case! Thanks for posting :).
Beej's guides are fantastic. He went to Chico State and I was happy to learn that his books were used for teaching network programming there when I attended.
Genuinely inspirational hacker. Feel's great to have had that experience at a state school.
Beej wrote the first computer science curriculum for Lambda School, which accounts for its initial breadth and depth. I taught there so I could learn CS, spending many hours to learn each class before teaching it.
One of my favorite modules he designed was on computer architecture. We simulated the CPU and registers in Python (they switched from C right before I got there), and each class extended our instruction set.
Teaching his modules taught me enough that I was able to start the MSCS program at UT Austin.
If you're trying to learn networking, better to start off with the foundational knowledge rather than jump straight into the code. Right now I'm going through the networking section on https://teachyourselfcs.com/ which is pretty good so far.
Contrary to popular belief, formal education wasn't useless here. All the foundational knowledge was covered in depth in my mandatory CS undergrad networking class. So this is a neat quick guide to all the stuff CS graduates already have learned once about but might've forgotten the details or don't know the hands on code.
I think there is (or was, depending on the institution) a disconnect between the curriculum and what skills students needed. For context, I was a tutor, instructional assistant, adjunct professor, and finally a curriculum advisor for the Los Rios Community College system and the University of California Davis Extension.
First, CS programs that went 100% into Java did not age well and most of these courses were taught in a way that led students on to believe this stuff was static and won’t really improve. The concept of new languages beyond what was available in circa 2006 was unimaginable.
Another quirk that a Java-centric education brought on was that low-level courses were often taught in Java and you ended up writing all your data structures coursework in Java taught by someone with a C background. But you couldn’t use C or anything sensible for this because universities promised students would know object oriented programming. This created a lot of disconnect and meaningless ritual around a paradigm that did not fit the language.
CS programs that succeeded understood that a foundational knowledge coupled with language flexibility. They had to understand that the specific languages did not matter so much, rather it should fit the paradigm and principles of the context in which it was written. Once students are able to think in terms of computer procedures, instructors can leverage this knowledge to introduce abstractions and a grand unification principle.
Funny I was just going over this again yesterday. Also didn't know he had one called Beej's Guide to Network Concepts[0], which has some really good explanations in it.
This is a fantastic guide and this was used in our programming lab at IIT Kharagpur. The lab was headed by an equally awesome instructor - Abhijit Das [1]
That's how Andrej Karpathy's video lectures are shaping up for me... I'm learning a lot about not just neural networks, but Python and Pytorch.
It's more or less magical when you find just the right conjugate match between tutorial material and your own learning style. I can see how Beej's Guide might do the same for someone new to both socket programming and C.
If you search Algolia for Beej you'll see his material has been on hacker news numerous times.