Semaphore Explained C#
Вставка
- Опубліковано 26 вер 2024
- In this Semaphore tutorial, I answer the questions: what is semaphore? what is semaphore used for? how to use semaphore? We go over a quick analogy to understand what the semaphore does and how to use it. We then look at a real world example of how it can be used to solve problems caused by asynchronous programming.
Patreon 🤝 / raw_coding
Courses 📚 learning.raw-c...
Shop 🛒 shop.raw-codin...
Discord 💬 / discord
Twitter 📣 / anton_t0shik
Twitch 🎥 / raw_coding
👉 Try Rider
www.jetbrains....
RD5K9-4TXXW-KMV3G-NYWSF-3ZSTP
Source: github.com/T0s...
#csharp
Thanks for the tutorial!
One note: when talking about semaphores, I think you missed the intended metaphor. Semaphore means "apparatus for signaling" and in this context it means railway semaphores, which you can think of "traffic lights for trains". There are two states for a railway semaphore: open and closed, so this is the intentional metaphor for the naming, it functions the same as your "gate" idea. =)
thank you, I didn't know
Great tutorial, thanks!
I imagine you’d want to move the release of the Semaphore to a “finally” block, otherwise those “gates” are permanently closed on an exception.
Yes, thank you for pointing out. :)
This channels needs more than a million subscribers ; wish i could subscribe a million times
Haha thanks man, we’ll get there ;)
Fantastically simple explanation, with both Hello World and real world examples. Short, sweet and simple - thank you.
Glad you liked it!
Well done young man - great stuff, and no edits. Impressive
My God!! You are the man!!! I had tried with many videos,codes,etc. Until I see your video and I did it succesfully in my first attempt!! Thank u man!!
Glad to hear )
thx from egypt, u saved my project because previously i was using deprecated thread methods such : suspend and resume which are have a lot of issues and not working properly. thx again
Glad to hear!
Best semaphore tutorial ever! I always struggled to understand it. After watching this, its crystal clear now! Thank you! The only thing I need to remember is the magic word - (Bill) GATEs!!!
Haha glad I could help :)
It's not that your network card is too slow to connect to Google 200 times at once, it's that your TCP/IP stack (and Google's too) has a set limit on how many unestablished connections can be buffered at once. In effect you're seeing this because Windows has its own "gate" and your socket requests start to block. The reason for this is to avoid allocating huge amounts of RAM to the TCP/IP driver in the event a program is stuck in a tight loop trying and failing to connect to something. Your actual NIC driver only knows about frames, not socket connections, and can blast data at upwards of 10Gbps. You're using HttpClient so the back-up may even be one layer higher (in http.sys) rather than in the native TCP stack itself.
Also you have 65535 theoretical TCP ports so you're also not running out of that.
I’ll take your word for it :D
This is the best channel! Thanks you so much for teaching all the stuff in an easy way!
Thank you for watching)
Clear explanation, thanks.
Thank you for watching!
Good and brief description. Excellent work. I like such format.
Cheers )
nice clear explanation and demo
Cheers
Finally the tutorial which resolved my issue. Thanks man😊
Thank you for watching
Excellent, just what I needed. Great job
Glad I could help
Nice explanation 👌
Cheers
Explains and sample are so good. Thanks for everything
Cheers :)
Great tutorial. Thanks a lot!
Cheers )
Thank you for the clarification. Like!
Thank you for watching
Nice little video :)
Thx for sharing.
Cheers
Awesome tutorial. I've now a better understanding about this topic
Thank you for your hard work and effort. I can see you enjoy teaching and I really enjoy learning. I hope to be as intelligent as you are and understand c# logic as the way you understand.
Anton fantastic explanation👍🏼👍🏼👍🏼👍🏼
Thank you ))
very concise and well explained! thanks brother!
Cheers))
Amazing video! Short, clean to the point.
Subscribed :)
Glad you liked it!
Self taught here your instructions are outstanding , I will be Grateful if you do tutorials about massaging concepts like rabbit MQ or apache kafka
At some point!!
Awesome explanation...
Cheers
@@RawCoding even others videos are too informative
Great tutorial !!!! Thanks buddy !!!
Thank you for watching)
Great tutorial! Thank you for taking the time to explain this so well.
Thank you for watching
Your explaination very clear. thank you for your video.
BTW, I really really miss LINQPad since I change to the Mac M1 :(
Mee too brother
Thanks for sharing! Great video. Thanks bro!
Thank you for watching;)
Thank you Anton, as always you discuss subjects that not a lot of people explain. If you are updating a financial account would it be better to use a semaphore or to use Rowversion and optimistic concurrency?
If you have 2 machines, semaphore is inside your app, at that point you would need a "distributed lock". Row version and retries is better.
nice one! - just a side note that SemaphoreSlim is a disposable object. so may be its a good idea to dispose it once used to save the resources.
If you need to dispose of it, dispose of it :)
Thanks!
Thank you for watching)
thanks for your gr8 video
thank you so much
Thank you for watching
Really love your videos! Very informative and well explained
Thank you for watching :)
Thanks, highly appreciated
Cheers
this was good!
Thank you for watching
Beautiful 🤩
Cheers
Great video!
Simple and elegant real-world scenario, thank you for the video!
I have came across a rather interesting problem where a block of code can't be allowed access by more than one thread, and that seemed like a perfect example for implementing the usage of the SemaphoreSlim class, however, I have an additional problem, my app will work on three nodes under a load-balancer and all three connect to the same database (the block of code checks and updates some data), and that will cause problems, this way the SemaphoreSlim will not help me in that situation.
What do you suggest?
You need a distributed lock or message passing
At around 7:32 I don’t understand how your calls aren’t blocking... the CallGoogle function awaits the _client call to google, then prints the result. So doesn’t that complete the call to google before returning? I would think each call to CallGoogle would be blocking. Is it because it returns a “task” that each “yield return” in CreateCalls doesn’t get blocked by that await?
(I’m not even sure I’m asking my question correctly. I’m a JavaScript programmer trying to learn C#. Is a task like a promise in JavaScript?)
That’s a good question. Yes Task is almost exactly like a promise. The reason they are not blocking is when I invoke CallGoogle it creates the Task and I don’t await on it, so the loop never stops and it quickly creates 200 tasks for me which all run asynchronously. All the 200 Tasks are awaited in a blocking manner by Task.WaitAll (there is a async version of this which is non blocking) in JavaScript you have the Promise.all function which is the non blocking version. Think of this like this we would take a list of strings and map them all to axios get requests, and the use Promise.all to await on them all so all requests are running asynchronously however in the “then” callback we just print the status code, internally between when the promise is resolved and the status code is printed the thread is awaiting, outside of the promise execution continues
Well played sir! :)
Ta
Thank you.
No thank you
Please, can you do one video on AutoResetEvent & ManualResetEvent ?
Thank you.
Thanks for the tutorial! Is there a default timeout on WaitAsync() or will that wait indefinitely?
Great
Thanks
I wish I could give you 1,000 likes!
Thank you))
'Try catch finally' would be more safely here with semafore releasing in finally section.
You are using the initialCount parameter, but there is also a maxCount parameter. Please explain what happens if you add maxCount?
I never used it tbf, but I’d imagine it’s the upper bound that you can go up to if you release more than you await
@@RawCoding That is, if, for example, initialCount = 5, maxCount = 10: This means that when 7 requests are executed simultaneously, 5 of them will be executed immediately, and two - a little later. But if there are 11 requests, then 5 will be executed immediately, 5 a little later, and 1 will not be executed?
If you lock and release for each request you’ll never end up past 5, so maximum 5 requests. I don’t actually know of any use cases for having a max count so can’t help with an example sorry.
What is Dump function and how to implement this?
It’s LinqPad specific
actually, I am working on my friends project it on c# and I want to popup a website from that software but we don't know how to change website links from server or database, we just need to change website link daily
Semaphore can't help with this sorry
Are you asking how to get the data from server or how to create links automatically
Probably then if I wanna have a gate at all endpoints I do some sort of midleware? nO CLUE how to do so.
I’d advise against middleware because that could pause your server. You want to bring this to your business logic level rather.
Wouldn't it be so much easier to just make that method not async so the call doesn't continue past and loop back for the result. Ending up with the the same outcome and less processing?
I guess if you use more than 1 it would make since this way to utilize more than 1 thread Now that i think about it and like he says as the developer you would need to tune it
Today most environments are multithreaded, blocking threads is a waste of resource. If we make the method non async that means we would have to block the thread, we want to avoid this.
Thanks for the video ....i like how you keep things simple. Im carious how this compare to other thread pool concepts for example the use of Parallel.Foreach where you specify Max # of threads? Also not sure about the yield use here, is it possible if we return each task at a time and the task finishes (hypothetically) before returning the next is possible the waitall will exit?
Sorry I don’t understand neither of your questions, I can’t compare it to Parallel.Foreach they are completely different, and the last part just confuses me
@@RawCoding sorry for the confusion. I guess my comparison to parallel.Foreach comes from the fact that like semaphoreslim you can limit how many threads can run asynchronously but maybe semaphore is how you can do it with async await concept. Regarding 2ed question i was trying to understand the use of keyword yield in "yield return callGoogle", doesnt that return each task one at a time to the Task.WaitAll in the Main function?
Oh for the 2nd question watch my generator video
I have question in another topic which is AsNoTracking and Tracking in entity core what is the difference ?
Not sure why you can’t google this question but here you go docs.microsoft.com/en-us/ef/core/querying/tracking
👍🏽
👍
Great. Statement that SemaphoreSlim is a newer version of Semaphore and it needs to be used in all the cases is incorrect, Semaphore is OS kernel mode object can be used for interprocess sync, SemaphoreSlim is a pretty much perpetual loop while(true) { if (!bResourceIsFree) continue; return; } => it's fast, but the cost is CPU usage
From what I see in the source code , waitAsync will create a task and store it in semaphore state as well as return it, release will queue that task to complete it on the thread pool. I don’t see any loop.
@@RawCoding Right. The loop is implemented in sync version. My bad
No worries, didn’t know the Semaphore was working on the kernel level tho thanks for that ;)
Is there anyway to know gate is occupied?
You can check the available count
I love you dude
so is semaphore and semaphore slim the same thing?
They do the same thing as in they are a semaphore, performance wise semaphores lim is better
Does semaphore slim work also between more assemblys, like mutex did?
Never used Mutex
500 запросов в гугл - верный способ, чтобы тебя реально "в гугле забанили" XD
Не там надо немного побольше постараться))
Good 👍, what tools you use for this coding?
LinqPad
@@RawCoding thank you
Thanks for sharing . Nice to know this.
Is it possible to make such implementation;
lets say semaphoreslim sm.=new semaphoreslim(20)
I want to put some sleep/timebreake after 20 request.
There’s no need to put sleep or break the semaphore is the break/sleep when you AwaitAync it will free the thread until the semaphore has more space.
@@RawCoding I want 5 seconds breake, between first 20 peope and second 20 people
You will need to do batch releases then.
I'm not gonna subscribe to this channel...No, I'm not...Damn it.
Welcome:D
Thx, nicely explained and demonstrated.
Thank you for watching!
Nice explanation👍
Cheers
Oh wow. I LOVE YOU.
Cheers
Thank you for the video. It is always nice to revisit some knowledge when you don't use it fairly often.
Thank you for watching))
thanks for this tutorial. how have you coded the dump() method?
It’s a method only available in LinqPad
@@RawCoding thanks !! I thought you have written an extension method.
Nope, that’s just LinqPad
This is excellent. A great way of explaining this. I struggled to understand Semaphores until now
Glad you like it )
Great explanation, the http requests example was very accurate
Genius
cheers :)
I don't see the point having your face in a programming video. That's not watchable.
Wow dude
Normal face)) I don't know English well, but thanks to facial expressions and gestures, everything is clear!