First time visitor. Senior dev that just wanted to see some different explanations on the topic in the context on how to best describe it. You explain in a concise and pedagogical way. You do this without all the the Bells and Whistles that usually comes with RUclips videos. I hesitate to search for explanations in video format because it often comes with too much distraction and ambition. Motivated or not, things I didn't want. This, however, really gets to the point in a straight forward and relaxing manner. Nice!
When I was looking for a video to explain the C# record, I knew I was in good hands. Always to the point, no big talk around it. Many greetings from Germany
That was like a super-concentrated C# learning syrup. Brilliant! Not sure I'll ever use records, but it's great to at least understand what they're for as I'm sure a use-case will pop up at some point. Great work - thank you!
I would absolutely love if this encourages people to put a bunch of (related) types into the same file so we don't get a bajilllion files just for simple types.
You should mention how and when a record will be allocated to the heap vs stack as this is important for performance and GC. Also, go into when and how copying of a record instance is a fill/deep copy, meaning no references are reused, and when it is a shallow copy, meaning value types are copied but reference types only have the pointer copied.
First of all, even if i point out a flaw, i really love your shorts, this video helped me alot in finding a simple way to explain this construct to my trainees and junior devs. The flaw im talking about starts at 1:50 of your video. There you are telling that you make a copy of the product by writing this: var product2 = product; But thats not true, you dont make a copy you just assign the pointer to another variable, you copied the pointer, not the object. You can test it by using this Code: var product2 = product; product.Description = "Description of Product"; Console.WriteLine($"Product1: {product.Description}"); Console.WriteLine($"Product2: {product2.Description}"); If i had a Copy, then only the first Console WriteLine should contain the Description string "Description of Product". But both does and thats because those variables actually do point to the same object and for that makes your explaination later with equals a bit weird.
Hi Shawn, could you give a video talking about the "readonly record struct"? The name seems a bit wired, and it is rarely mentioned. What is its application scenario?
This is probably a good fit, but to me the best use case for records is multithreading. Since you can’t edit them after the fact, it forces you to use them in a thread-safe manner.
Not sure why that would be. Can you explain? I'm curious. Difference between struct and record is inheritance (though likely not that useful but good to know).
@@swildermuth One is on the heap the other isn't. Correct usage of structs instead of classes in high volume systems minimizes garbage collection. It's unfortunate that you can't inherit or at least say "implement identical properties" of a struct, but when using protobuf of messagepack or even graphql which generate DTOs from other formats, that doesn't really matter, and the perf improvement is pretty nuts.
Thanks for doing these insighful coding shorts. I noticed you use a shortcut to run the 'dotnet.exe' -was it a case of just making a copy and naming it 'n.exe' (in the same folder)?
You can also mutate the record back onto itself e.g. product = product with { Price = 1.99m } I use this in Fluxor where the current state is passed into a reducer method.
That's aggressive. Sure, struct with aligned fields is faster, but to what end? Faster != better code in many cases. Sure, you're building a real-time system, every cycle counts; but the ability to do immutable data can make code better and more thread safe. If you're optimizing for cycles, but most of your work is on the network, your time savings is swallowed up in the aggregate. Let me guess, non-wankers write in Rust or Go?
First time visitor. Senior dev that just wanted to see some different explanations on the topic in the context on how to best describe it.
You explain in a concise and pedagogical way. You do this without all the the Bells and Whistles that usually comes with RUclips videos. I hesitate to search for explanations in video format because it often comes with too much distraction and ambition. Motivated or not, things I didn't want. This, however, really gets to the point in a straight forward and relaxing manner. Nice!
When I was looking for a video to explain the C# record, I knew I was in good hands. Always to the point, no big talk around it. Many greetings from Germany
Glad you liked it!
Sweet. That makes sense. Didn't really see how record was useful, but this makes sense. Thanks!
No problem!
That was like a super-concentrated C# learning syrup. Brilliant!
Not sure I'll ever use records, but it's great to at least understand what they're for as I'm sure a use-case will pop up at some point.
Great work - thank you!
Thanks!
I would absolutely love if this encourages people to put a bunch of (related) types into the same file so we don't get a bajilllion files just for simple types.
You have an excellent concise teaching style.
Short and simple! Thanks!
You bet!
this was the clearest explanation video I've ever seen in a while. GJ!
Glad it helped!
The best explanation you can ask for
Thanks. I am glad it helped.
Brilliant video, that was so concise and informative, thank you.
Glad it was clear!
You should mention how and when a record will be allocated to the heap vs stack as this is important for performance and GC. Also, go into when and how copying of a record instance is a fill/deep copy, meaning no references are reused, and when it is a shallow copy, meaning value types are copied but reference types only have the pointer copied.
this is standard knowledge so there’s no need to re-explain this.
Amazing explanation and demonstration keep it up!
Thanks, will do!
Video and audio quality are ridiculously good. Need S Dubs in the highest def possible
S Dubs?
@@swildermuth Your initials
First of all, even if i point out a flaw, i really love your shorts, this video helped me alot in finding a simple way to explain this construct to my trainees and junior devs.
The flaw im talking about starts at 1:50 of your video.
There you are telling that you make a copy of the product by writing this:
var product2 = product;
But thats not true, you dont make a copy you just assign the pointer to another variable, you copied the pointer, not the object.
You can test it by using this Code:
var product2 = product;
product.Description = "Description of Product";
Console.WriteLine($"Product1: {product.Description}");
Console.WriteLine($"Product2: {product2.Description}");
If i had a Copy, then only the first Console WriteLine should contain the Description string "Description of Product".
But both does and thats because those variables actually do point to the same object and for that makes your explaination later with equals a bit weird.
I didn't mean to imply that. My bad.
@@swildermuththanks for the response
great explanation! i've struggled with how classes and recs are different, so this really helped a lot.
Great explanation, thanks!
Good video, and ur voice is so relaxing sir.
Thank you kindly!
You didn't really say why we should use records but thanks anyway.
At 7:55 he discusses when he uses them and (perhaps not exhaustively), why.
I love your short but very informative videos! You just got a new subscriber!
Thank you, you are great
How would a collection of objects work inside a record? Would it still be value type?
The container for the collection would be, but, obviously, not the objects themselves. So very little savings.
Hi Shawn, could you give a video talking about the "readonly record struct"? The name seems a bit wired, and it is rarely mentioned. What is its application scenario?
It's well documented on the Microsoft Learn site.
Cool video 👌
Thanks for this!
But what's your take on creating DTOs with records instead of with classes?
This is probably a good fit, but to me the best use case for records is multithreading. Since you can’t edit them after the fact, it forces you to use them in a thread-safe manner.
What I really want to do is make all of my DTOs readonly record struct. But it doesn't work in a number of cases including [AsParameter].
Not sure why that would be. Can you explain? I'm curious. Difference between struct and record is inheritance (though likely not that useful but good to know).
@@swildermuth One is on the heap the other isn't. Correct usage of structs instead of classes in high volume systems minimizes garbage collection.
It's unfortunate that you can't inherit or at least say "implement identical properties" of a struct, but when using protobuf of messagepack or even graphql which generate DTOs from other formats, that doesn't really matter, and the perf improvement is pretty nuts.
Thanks Shawn
yw
How is that last bool method you wrote a part of the record above? What indicates that?
It is just like a class declaration, you can add members (in this case a method) in the body of the record. I think that's what you're asking.
Thanks this was great.
Glad you enjoyed it!
how do I convert records into classes and vice versa?
i use chatgpt
@@ghevisartor6005 😆👍
Thanks for doing these insighful coding shorts. I noticed you use a shortcut to run the 'dotnet.exe' -was it a case of just making a copy and naming it 'n.exe' (in the same folder)?
I got it from here: github.com/faniereynders/dotnet-sdk-helpers
I renamed their dot.cmd to n.cmd
@@swildermuth thank you
You can also mutate the record back onto itself e.g. product = product with { Price = 1.99m }
I use this in Fluxor where the current state is passed into a reducer method.
Technically true, but it is new copy (a new reference). Right?
@@swildermuth yeah correct, it creates a new copy and assigns it back to the variable.
helpful👍🔥
Glad to hear that
thanks my amercian bro
You're welcome
Are you the voice of Lester in GTA5?
I wish!
nice explication man; but nothing will be fast as a struct with aligned fields; my feeling is that C# is becoming a language for wankers sadly
That's aggressive. Sure, struct with aligned fields is faster, but to what end? Faster != better code in many cases. Sure, you're building a real-time system, every cycle counts; but the ability to do immutable data can make code better and more thread safe. If you're optimizing for cycles, but most of your work is on the network, your time savings is swallowed up in the aggregate. Let me guess, non-wankers write in Rust or Go?