Hi Hussein I just wanted to say that I got an admission offer from Technical University of Dresden for masters in Distributed Systems Engineering. And it wouldn't have been possible without your videos. I've been following you since the last couple of years and your content has been extremely helpful and thought provoking. I wouldn't have been here without you. Thanks and never stop making free content like this. ❤️
Hussein's pacing, expressions and intonation add so much context to these complex topics. 13:14 - summarizing layer 4 proxying with a single gesture and noise - I understood instantly.
Yo this video was awesome. Web has evolved so much that it’s hard to understand how we landed on the ‘latest solution’. Knowing the history helps answer the ‘why’ for the technology. Thank you
11:35 and this why i love Postgres, has so many capabilities that i love tinkering with and i dont feel limited but know that if i need this feature from sql its probably available.
Awesome, my friend and I had to use WebRTC in order to have car game with decent UDP support for fast data transfer like playerPostion, while all other game logic was with WebSockets reliable (match start, match end, etc). WebRTC was pain in the ass so I am glad to hear there are some new protocols that as you said gives us control how we want our data be sent/received and how we deal with that on app level.
Unfortunately, people tend to give up control too easily. I understand that when dealing with simple tasks, it can be tempting to delegate to black-box solutions without much consideration. However, this mindset can foster a dangerous culture. While working with raw streams or low level APIs may require more code and can increase the risk of mistakes, it also presents valuable learning opportunities. Engaging with low-level APIs, streams, and similar concepts can ultimately make you a better developer, enhancing your skills and understanding.
best is to have both APIs, high level and low-level. So when someone needs to build something simple quickly, high-level will do... The alternative is: having a low-level API and a good open source/free software libraries/framework that implements a high-level API.
I kind of agree. I do think that people should be more willing to familiarize themselves with lower level concepts, rather than just treating them like black boxes as you put it. Even if you have no desire to work with code at that level, being more familiar with the mechanics of it can help you utilize it better, and avoid bad use cases. However in a production environment, trying to rebuild everything from scratch without first having experience on the higher level is a recipe for disaster, and a great example of premature optimization (regardless of whether you're trying to optimize for execution time, memory usage, stability, scalability, etc). I would also say this would be much easier to get people to do if developers would stop writing documentation which is ostensibly for 3rd parties, but in fact only makes sense to people who are already intimately familiar with it. This whole hazing routine of "I-had-to-suffer-reading-shit-documentation-so-you-do-to" really needs to stop.
I'm guessing WASM has something to do with this. With the required low level primitives, WASM would get portability between a real OS and just a web browser
Small nitpick, you seem to have mixed up unreliable with unordered, when describing datagrams (which are obviously both). In practice the difference is pretty small, but if someone uses datagrams for online multiplayer games, then it will matter.
Hussein Nasser I was wondering why, now that we have IPv6, which can cover all devices in the world, each device has its own IP address so we don't have to design new bidirectional protocols such as sockets
Yes, I get that, but I mean that if each device has its own IP, we don't need bidirectional protocols since each device acts as a server, and the server may obtain the IP addresses of all linked devices and transmit them straight to a specific device. because the main purpose of the bidirectional protocol is to allow the server to send data back to the client and keep the connection alive
@@saeedalobidi4195 oh i see, i think because bussiness and permit management. thats why each client still dont get own public ip. it more easy and scallable to use NAT, one public ip used by several client by assign private ip.
I know I should get the answer of this by profiling, but wanted to know from all other fellow engineers here, Will WebTransport protocol (over HTTP/2) be ideal for a high throughput usecase (like sending the LTP of a stock or cryptocurrency)?
Hello Hussein, great video but i have one question at 13:10 you mentiontioned something like layer-7 and layer-4, what does that mean, I many sound silly asking this question but I only have 3 years of experience with web development.
We have epic WebRTC failure, where great idea about P2P connection was killed by NAT. Will WebTransport can be useful in real world with all this proxy, CDN e.t.c and wild zoo of browsers at client side? I see somone still support they project for IE.
Hey Hussein, I love your content but sometimes it would be nice if these were summarized into 5 minutes or less. My short attention span gets the best of me after that point. Just my unsolicited opinion and hopefully you take it with positive intent. Thanks for the content!
WebTransport is available on both HTTP/2 and HTTP/3. QUIC is an HTTP/3 thing, basically building around head-of-line issues with TCP by using UDP instead.
@@JetpackBattle-lc7ob I believe the BSD Sockets library let’s you use TCP and UDP. Perhaps I shouldn’t have been so specific citing TCP. I never completely understood why Websockets is a thing when actual sockets are available as an option, and have been a thing since the inception of the Internet.
hm. Looks good, i am intrigued by this prop of the transport "draning", I have not seen that mechanism often. I still wonder where it stops, about the networking APIs. There has been several, it is not like they did not have success, but we would not be here if something was not off.
