0:00 - Intro 0:57 - Getting started: demo console app 2:16 - How to use Records 3:45 - Records: declared and explained 6:55 - Record features 13:00 - Records vs. Class: ToString() override 17:04 - Records vs. Class: compare objects 20:49 - Records vs. Class: compare object reference 22:10 - Records vs. Class: compare object == 24:31 - Records vs. Class: object Hash Code 27:10 - Records vs. Class: Recap 28:13 - Records vs. Class: Deconstructor 31:32 - Records vs. Class: Record Object copy 34:30 - About Records 36:19 - Record property access modifier 40:12 - Other Properties in Records 42:53 - Methods in Records 43:56 - Full Properties in Records 47:33 - Recap 49:49 - Record inheritance 52:57 - *Why Records?* 53:53 - /*Benefits*/ 57:45 - /*When to use Records*/ 1:05:10 - /*When not to use Records*/ 1:14:31 - Summary and concluding remarks (+ bonus: top level call args in C# 9) You forgot the time codes ;)
Superman Tim Corey, could not resist getting into yer slipstream. Really great to have you on board. Thanks for yer forte which appears to be a mind boggling easy way to get into web development.
This is amazing; I was recalling concepts from Java Script and Java. It sounds like the smart people at Microsoft worked so hard to make available capabilities that make coding a lot easier. Things that in other languages you just have to remember how to do. But now in C# you have specific methods with specific results. I love it. And you explained it so well as always. Nothing against other languages I love them all, and wish I had time to learn more about them. But I am so glad I decided to learn C# and other Microsoft technologies.
Thanks for such a detailed explanation, Tim. You are the best at explaining complex concepts and having the perfect analogies to better understand them. The house and address analogy for class addresses was brilliant.
Hi Tim, I'm going to Copenhagen academy in IT learning and I'm doing web programming in server, all my fellow students are more interested in your videos than in our teacher, I post all your free videos on our forum, well done Tim, I also buy your courses and learn a hell of a lot from them many greetings Knud Nielsen Denmark
@@IAmTimCorey I have recommend you to person there start at university study at programming and they are thankfull for your videos, and I wish your lection in Blazor to christmas it's great's
Awesome material, and a great explanation! For those trying it at home, there are a couple of syntax simplifications you can use: public record Record2(string FirstName, string LastName) { internal string _firstName = FirstName; internal string FirstName => _firstName.Substring(0, 1); public string FullName => $"{ FirstName } { LastName }"; }
Thanks Tim, I wish you work for MS because your explanations are always crystal clear, unlike MSDN articles which sometime looks like "how to draw Owl" meme.
Thanks for another great video Tim! At 10:58, what I have also learned from the 'Ctrl + .' boiler plate reduction is you can type your 'firstName' and 'lastName' parameters before typing them as properties. When you 'Ctrl + .' it will give you the option to create the parameters as properties or fields. I have found this helpful in creating DTOs and dependency injection.
10:30 There's actually an even faster way to do it, you select all the properties that you want in your constructor, then hit alt + enter, there's an option that says generate constructor, it will generate all the constructor for you. You don't have to write anything. I use visual studio 2022 but idk if you can use it in 2019 and below.
At 10:38 of the video when you use the trick to initialize the field value from the constructor, it's even faster if you do Control + Dot on the private field instead of on the parameters of the constructor. In other words have an emty constructor then do Control+Dot on the field and select the option to "Initialize from Constructor". Long video but great thanks Tim
Once I saw a Python talk (I also code in python) and in that talk one thing stuck with me: "Every feature [...] has at least one or two very clear and unambiguous mental models [...] for where is actually useful." The talk was about some python features, like metaclasses, but I found that this thought process can be applied for most of programming in general. I believe is more importante to understand a feature in a abstract level (its mental model/purpose) rather then all the technicalities that you can look up in the documentation. Same goes for records. What is a record, in real life? According to a dictionary: "1. a thing constituting a piece of evidence about the past, especially an account kept in writing or some other permanent form." It stores information about the past. We can't change the past, so we shouldn't change records. So there you go, records are for things that we cannot, or should not, change over time. Those mental models really help me get a hold of when to use certain features.
That's why I believe set shouldn't be valid for records, only init. But I think they've focused too much on the value equality and other stuff and forgot about this
Hey Tim, Good stuff, thankyou. Call me crazy but I could swear you said multiple times "location on disk" when differentiating between a record and class You said it so many with confidence it makes me think i'm missing something. (one of us is going crazy :))
I'm not sure if I said disk but I did talk about the physical location of the class instance. That would be the location in memory that is carved out to store your class instance. With a record, when you copy it using the with statement, you are duplicating all of the values into new memory storage. With a class, when you change a value, you are just changing that specific memory value to be something different.
Hi Tim great video as always. I'm watching all yours video and very often you are copying line of the code using copy and paste. I have found very useful to put the cursor on the line that I want to copy and press CTRL+D to copy whole line and paste below. Thank you for sharing your knowledge.
What Luke S just said also works if you select a whole block (say, a method you want copied to make an async version). Oh, and that Ctrl+dot thing on the constructor, to initialize your properties? Try selecting one (or more) properties and hitting Ctrl+dot. It'll ask if you want to initialize it in a constructor... it'll even create the constructor for you if you don't have one yet! :D Saves much typing.
Great video Tim! Very clear like all your others one. Just one small observation! public record User1(int Id, string FirstName, string LastName) : Record1(FirstName, LastName); I think that this line is a bit of confusing because it's the same thing of write: public record User1(int Id, string FirstName, string LastName); The example that would have made the most sense would have been to inherit from Record2 in order to show more clearly that the methods/properties/etc will be inherited too
Your line isn't the same as mine - Record1 has a constructor and we need to pass data into that constructor to satisfy it. So, when we inherit from Record1, we also tell it which two properties to send into Record1's constructor.
Hello Tim, I listened to the whole thing. Very informative, but one last bit you said around 1:13:15 tripped me. You said the record clone is making a shallow copy. But how so? We didn't have any reference types as any properties in the record to begin with. What exactly is shallow copied when you clone a record? Did you mean the reference to the record itself is reused on a clone?
In C# strings are reference types, so in this case when he is creating a new record from a copy of the original and replacing the FirstName value the new instance re-uses the LastName string. This works nicely with multiple levels of records since each instance is immutable so keeping references all over the place as you change only parts of a sub-record is a great performance gain without worrying about state changing underneath you.
45:26 Actually, C# will handle way better that we will, TBH. I like the idea of having immutable data philosophy and being closer and closer to functional paradigms in this C# new era. In my opinion, record types comes with lots of extra stuff which is only shadowing what they should be, which is *immutable data structures* . The idea of redeclaring properties and the ability to declare methods makes records less responsable. *Data structures must store data* , nothing more, nothing else. If any system needs to hide some properties from a certain data structure or maybe remapping some of their properties, then the system must do it by itself (by implementing filters, mappers, handlers, etc), not the actual data structure.
5:58 - Records act like value types, but they are not 7:44 - Meaning of "init" keyword 8:03 - Records are immutable 17:22 - Records equality 21:04 - Reference equality 23:27 - Equals() and == are not the same 28:19 - Record deconstructor 32:00 - Copying and modifying records
Great video! Thoroughly enjoyable. I would just like to mention a couple of more use cases that records may be useful for. One thing I especially like about them, is how they can be declared with very little additional boiler plate code. They can be declared almost completely in domain language and they are compact, making them easy to read for non-tech oriented domain experts. They can be used as a tool for defining the domain knowledge in code, and being understandable between domain experts and developers! Another thing they can be really useful for is when combined with the new pattern matching features (Especially switch expressions) - They can be used as a reasonable substitute for discriminated unions, as they are able to be derived from - This, combined with pattern matching is a very powerful combination, and I think would probably warrant its own video. I loved the explanations, and I dont know why MS made the choice to allow mutability.
26:48 very dangerous advice here: do not expect 2 objects producing the same hashcode to be equal, that's not a guarantee of the GetHashCode() contract
I'm using records with Dapper now, and it's working out just fine. Then, I have a different approach to data that needs to be changed/updated: I use the copy feature of records. That makes it distinctly clear exactly which fields will be changed. That's a lot more readable than using classes and having to copy every property. Using a record copy exposes clearly what my intentions are, and leaves much less space for bugs to hide in. All my records are still immutable, my code is brief and focused on what I do. it's good :)
@@IAmTimCorey is there a way to capture all of these after video creation knowledge into the description? I knew someone would have an answer to this but I had to skim past several comments to get an answer.
As the Model or the ViewModel? If the model then it could be as long as you don't manipulate the data coming in or going out. It would not really work as the ViewModel because the VM inherits from a class and it allows for data change.
Very informative. I did a couple of projects using Akka.Net. Since the objects passed around by Akka must be immutable, records are an excellent way to encapsulate the Akka message data and you don't have to worry about accidentally changing your object.
I certainly found the use case for records in our project. For serverside caching, we calculate the SHA256 hashcode of the requests to identify if the request getting repeated, I see records as a potential replacement for classes there as we don't need to change the incoming requests and want the same hashcode every time the same request comes. Scope to get rid of SHA256.
I would add that records are useful when writing more functional code (c# seems to be migrating to be a hybrid language) and if you are using records it should be records and immutability all the way down. A developer can run into issues if a record has a class that is mutable making the record shallow immutable. They do a solid 85% of the work for us but you can still shot yourself in the foot.
After watching I was dying to check this out: static void Main() { var r1 = new Record1(); var r2 = new Record1(); Debug.Print($"r1: {r1}"); Debug.Print($"r2: {r2}"); Debug.Print($"r1 == r2: {r1 == r2}"); } public record Record1() { public int RandomNumber => new Random().Next(0, 100); } What do you think the last Debug.Print will produce? Take your guess! :D P.S. Thank you Tim for another great video!
Thanks for all the content you always deliver. I made huge progress in C# over the last year and your videos played a big part on that :) I have two examples where I'm still not sure what I would use: 1. A Coordinate. Usually it was a prime example for a struct. But with all the features of a record, i would tend to create it as a record now. The easy comparison, an easy destructor, immutability and also the ability to have for example a 3DCoordinate inherit from Coordinate. But this makes me wonder, what are cases to use a struct then? (I almost never used structs in the past, do i don't know that much about those) 2. Let's say we have a simple n:1 relationship between city and person (inhabitants) and none of the cities properties will ever change. Just the objects in the Collection for inhabitants. Would it make sense to use a record here, if i want to guarantee immutability?
10:47 this also works vice versa, when you've defined your members like FirstName and LastName already, .-space on an member and you should get the option to create an constructor with parameters. I believe you can even choose which parameters. While ctor only gives blank constructor, this gives you parameterized constructors.
Thank you so much for sharing your vast knowledge with us, Mr. Corey! I think I am going to use this records for my DTO objects. This should work fine I guess because I don’t think they (DTOs) should be mutable.
Thank you so much for the awesome video! Everything was very well explained! Only thing I was still wondering was if Records can inherit Interfaces or not. I searched the web and I see that they can! :)
"Personality, I wish that classes would do this". Same. For a lot of my classes I simply override ToString() to output a json representation of the class. For the className I simply declare it as a string.
Yes, but then User1 would not be a Record1, so we could not assign a User1 object to a Record1 object (not a big deal, just something to note). If we had inherited from Record2, there would be more implications because of the custom code in Record2 that we would be inheriting.
@@IAmTimCorey I figured, thanks for the clarification. I might give records a go for a project at work. but I am running into issues with deserializing from JSON into a record with the JsonPropertyName annotation. Are records suppose to work with annotations? I don't get error messages, but my record comes up with no values when i print it, if i take the annotations off it works, but i have to match the Json names exactly, which causes a problem property name casing.
@IAmTimCorey I figured out doing something like this worked out public record PlanetRecord(string Name ... rest of the properties) { [JsonPropertyName("name")] public string Name { get; init; } = Name; ... rest of the properties }
Hey Tim, love this video. Totally unrelated but I wanted to let you know that if you do ctrl + enter at the end of a line, it will copy it and put in the next line. I saw you doing a manual copy paste :) Nice shortcut when you want to copy and paste a line. Thanks again for the vid.
I know this is old but once you override the properties you have to make a constructor and deconstructor for the overridden value for it to retain its original functionality, the reason is you changed it to a non-positional record, similar to how a list works. You told the record I need this to behave differently so it pulled the value from its position does the changes you declare then pops it back at the end as it moves in order. So in your example Last name moves to the front of the order and the changed Firstname moves to a new position behind it. The reason you can use { Get; Set; } is because you need to be able to use them for a public record struct as a struct cant be {get; init; } your case in Do not do is an exploitation of this behavior by hybridizing a loophole. When making the record a struct it takes on the behavior of a value type not a reference type removing the need of immutable. In fact as an example: public record struct Point(float x, float y) //initializes as public struct Point { public float X { get; set; } public float Y { get; set; } public Point(float x, float y) { X = x; Y = y; } } unlike public record Point(float x, float y) //initializes as public class Point { public float X { get; init; } public float Y { get; init; } public Point(float x, float y) { X = x; Y = y; } }
@ IAmTimCorey I like using record's when working with Json (System.Text.Json). I write a record that describe my format, then I serialize and desterilize it when needed. Also, I like using the custom Deconstruct void in my record to do Pattern Matching.(Fun Stuff)
Great video as always! just got around to learning this. 1 quick question - where would you define those records? in classes, we usually define each class in its own file is this the same for records?
Hello, Tim. Thank you for this lesson. It was very interesting and useful. But I have a question. As I know all types for collections are reference types. And I can't use List or Array type for property in record because their hash codes will be different for different objects. So how can I work with record type that has to contain collection?
@@guiorgy, or you can write your own collection type and override GetHashCode and Equals methods. Also there are some types that implement IReadonlyCollection, but I didn't investigate how they work. May be these types can garantee that hash codes will be the same for same collections. Need to check it.
@@levkurbanov431 From a quick Google, seems that readonly collection also don't work. You'd have to extend collections with value semantic Equals and hashes, and if not careful, something else might break, like Json deserialization. In general, reference types without value type semantics should be avoided in Records, I guess. Source: stackoverflow /a/69840868
the equality checks make records compelling for me. I worked on a project where we kept multiple versions of documents in mongo, so on ingress (creation) we would find an existing, and compare so as to create a new version. Needless to say, we had to override gethashcode based on each of the properties, and then write/maintain equality tests for each such class.
@8:00 you are comparing init properties to the record and that your previous video etc. and that you can set those values via curly braces during construction. Is there a way to do that with Records? When I am trying it, I am required to use the constructor. Is there some option I missed in your video to allow me to use curly braces construction?
As someone new to C#, but experienced in a number of other frameworks, paradigms, and languages. You spoke in the use cases about the use of these immutable types (Records) when Parallel processing or reference data is being loaded. I can see how you can hard code a Record, but when you have a Data Source (such as the output of an API in JSON or a set of "Objects" from an ORM to your Database via .NET), I can see how I can easily deal with the ingestion of those into a Class, but can Records be the subject of an ORM easily or do I have to unravel things from a class that duplicates the content that I'm interested in -- inheritance from Class to Record would seem to be a logical way to bridge this. At the end you use composition with the Class having a Record, can that be the subject of an Object Relational Mapping without a ton of manual assignments? Awesome quality and detailed dive into Records, I'll be on the look out for other subjects on TH-cam and take a look at the courses you mentioned. I may have typed too soon. You mentioned "Dapper" in your Don't section and EF is not working yet. You see I would like to have an Add/Change/Delete C# App to maintain my massive DataSet and then have Records in my Analysis App with a way to share definition of the Database Access to help with changes between the two. Anyhow, any insight you can impart would get great or another Video would be even better :^)
Well, my video introducing Dapper will help you: th-cam.com/video/Et2khGnrIqc/w-d-xo.html (I also have an Advanced Dapper video if needed). I'll be showing off Records more soon (probably in next week's video, for example).
Thanks!
Thanks for the support!
Thank you!
0:00 - Intro
0:57 - Getting started: demo console app
2:16 - How to use Records
3:45 - Records: declared and explained
6:55 - Record features
13:00 - Records vs. Class: ToString() override
17:04 - Records vs. Class: compare objects
20:49 - Records vs. Class: compare object reference
22:10 - Records vs. Class: compare object ==
24:31 - Records vs. Class: object Hash Code
27:10 - Records vs. Class: Recap
28:13 - Records vs. Class: Deconstructor
31:32 - Records vs. Class: Record Object copy
34:30 - About Records
36:19 - Record property access modifier
40:12 - Other Properties in Records
42:53 - Methods in Records
43:56 - Full Properties in Records
47:33 - Recap
49:49 - Record inheritance
52:57 - *Why Records?*
53:53 - /*Benefits*/
57:45 - /*When to use Records*/
1:05:10 - /*When not to use Records*/
1:14:31 - Summary and concluding remarks (+ bonus: top level call args in C# 9)
You forgot the time codes ;)
Yeah. U took a LOT of time to do it, when Tim already knew the times...
Thanks! MUCH appreciated
Thank you Tim, seriously this is the first time I know about Records in .Net core.
Please give us all updates in C# as much as you can.
Records are brand new, so you haven't been missing anything. They were just released in November 2020.
Superman Tim Corey, could not resist getting into yer slipstream. Really great to have you on board. Thanks for yer forte which appears to be a mind boggling easy way to get into web development.
Thank you for the support!
Great, in my humble opinion, i think you deserve the c# crown
Thank you!
This is amazing; I was recalling concepts from Java Script and Java. It sounds like the smart people at Microsoft worked so hard to make available capabilities that make coding a lot easier. Things that in other languages you just have to remember how to do. But now in C# you have specific methods with specific results. I love it. And you explained it so well as always. Nothing against other languages I love them all, and wish I had time to learn more about them. But I am so glad I decided to learn C# and other Microsoft technologies.
Thanks for watching and sharing!
Thanks for such a detailed explanation, Tim. You are the best at explaining complex concepts and having the perfect analogies to better understand them. The house and address analogy for class addresses was brilliant.
You're very welcome!
Hi Tim, I'm going to Copenhagen academy in IT learning and I'm doing web programming in server, all my fellow students are more interested in your videos than in our teacher, I post all your free videos on our forum, well done Tim, I also buy your courses and learn a hell of a lot from them many greetings Knud Nielsen Denmark
Awesome! I am glad they have been so helpful.
@@IAmTimCorey I have recommend you to person there start at university study at programming and they are thankfull for your videos, and I wish your lection in Blazor to christmas it's great's
Awesome material, and a great explanation!
For those trying it at home, there are a couple of syntax simplifications you can use:
public record Record2(string FirstName, string LastName)
{
internal string _firstName = FirstName;
internal string FirstName => _firstName.Substring(0, 1);
public string FullName => $"{ FirstName } { LastName }";
}
Thanks for the tips!
Thanks Tim, I wish you work for MS because your explanations are always crystal clear, unlike MSDN articles which sometime looks like "how to draw Owl" meme.
You are very welcome.
Sir, I love the way you teach.
Thanks!
Thank you so much Tim, records definitely needed their own video.
Very welcome
Thank you very much for this valuable knowledge.
My pleasure
I will definitely reference this video to my colleagues
34:43 Difference between struct and record
36:02 Records can have methods and can inherit from other records
52:59 Why we need records
Great educational presentation with high quality! And the chapter breakdown makes it easy to jump between the parts you find most useful.
Glad it was helpful! Give Ralfs thanks for the timestamps. He has done dozens of videos as a way to give back to this great community of viewers.
Thank you so much Tim, no matter what level a developer is, ur uploads are great !
Glad you like them!
Thank you for the big fonts.
You are welcome.
Thanks for another great video Tim! At 10:58, what I have also learned from the 'Ctrl + .' boiler plate reduction is you can type your 'firstName' and 'lastName' parameters before typing them as properties. When you 'Ctrl + .' it will give you the option to create the parameters as properties or fields. I have found this helpful in creating DTOs and dependency injection.
Great tip!
Thanks Tim, I knew I could only get vivid description of the updates from you. Looking forward to the Blazor updates.
More to come!
Thank you very much for this valuable knowledge.
Records - A value-added to C#
You are welcome!
10:30 There's actually an even faster way to do it, you select all the properties that you want in your constructor, then hit alt + enter, there's an option that says generate constructor, it will generate all the constructor for you. You don't have to write anything. I use visual studio 2022 but idk if you can use it in 2019 and below.
Thanks for sharing!
At 10:38 of the video when you use the trick to initialize the field value from the constructor, it's even faster if you do Control + Dot on the private field instead of on the parameters of the constructor. In other words have an emty constructor then do Control+Dot on the field and select the option to "Initialize from Constructor". Long video but great thanks Tim
Yep, that works too.
Once I saw a Python talk (I also code in python) and in that talk one thing stuck with me:
"Every feature [...] has at least one or two very clear and unambiguous mental models [...] for where is actually useful."
The talk was about some python features, like metaclasses, but I found that this thought process can be applied for most of programming in general. I believe is more importante to understand a feature in a abstract level (its mental model/purpose) rather then all the technicalities that you can look up in the documentation. Same goes for records. What is a record, in real life? According to a dictionary:
"1. a thing constituting a piece of evidence about the past, especially an account kept in writing or some other permanent form."
It stores information about the past. We can't change the past, so we shouldn't change records. So there you go, records are for things that we cannot, or should not, change over time. Those mental models really help me get a hold of when to use certain features.
That's why I believe set shouldn't be valid for records, only init.
But I think they've focused too much on the value equality and other stuff and forgot about this
You don’t update the number of victims when the serial killer strikes again?
I think we need to update his record.
😎
There'll be a new record of a new crime. No past records would change
Nicely done, thanks for sharing
More people need to watch this video!! I was having a hard time understanding records and this cleared everything up!
Excellent!
Thank you Tim for this awesome content. The comment section is educational as well.
You are welcome.
Hey Tim, Good stuff, thankyou. Call me crazy but I could swear you said multiple times "location on disk" when differentiating between a record and class You said it so many with confidence it makes me think i'm missing something. (one of us is going crazy :))
I'm not sure if I said disk but I did talk about the physical location of the class instance. That would be the location in memory that is carved out to store your class instance. With a record, when you copy it using the with statement, you are duplicating all of the values into new memory storage. With a class, when you change a value, you are just changing that specific memory value to be something different.
@@IAmTimCorey Yes, i know.... but you said "location on disk like 5 times". No worries. Will be interesting to see how this new Record thingy is used.
@@IAmTimCorey Well... perhaps not 5 times, but for sure 2 or 3 at least. You can hear an example of this at 24:11.
I was also wondering about this. Glad Tim replied to clarify it wasn't intentional.
Great video and Thanks for the detailed explanation Tim....
You bet
You forgot your fav. Putting records in lists.
It works.
Thanks for the video :)
Also 11:57
ctrl+d.
Yeah, I always forget the line duplication shortcut. I only use it in demos.
Thanks Tim, this is a great video!
Glad you liked it!
बहुत ही अच्छा वीडियो टीम जी आप ऐसा ही ज्ञानपुर्वक वीडियो हमारे लिए लाते रहिए । हमारी दुआ आपके साथ है । धन्यवाद 🙏
बहुत-बहुत धन्यवाद
Tim Sharp Corey 👍👌
Thanks
Excellente video indeed. Everything was thoroughly explained. Thank you!
You are welcome.
Perfect timing, as I head for my lunchtime coffee !
Worried about 1:19:44 long coffee break 😀😂 just joking
Yes, it does sound like a LOT of coffee.
Thank you Tim! This is a high quality Knowledge.!
My pleasure!
Absolutely top content Tim!! More power to you
Thank you!
Hi Tim great video as always. I'm watching all yours video and very often you are copying line of the code using copy and paste. I have found very useful to put the cursor on the line that I want to copy and press CTRL+D to copy whole line and paste below. Thank you for sharing your knowledge.
What Luke S just said also works if you select a whole block (say, a method you want copied to make an async version).
Oh, and that Ctrl+dot thing on the constructor, to initialize your properties? Try selecting one (or more) properties and hitting Ctrl+dot. It'll ask if you want to initialize it in a constructor... it'll even create the constructor for you if you don't have one yet! :D
Saves much typing.
Yep, I just forget that shortcut since I use it so infrequently (mostly just for demos).
Great video Tim! Very clear like all your others one.
Just one small observation!
public record User1(int Id, string FirstName, string LastName) : Record1(FirstName, LastName);
I think that this line is a bit of confusing because it's the same thing of write:
public record User1(int Id, string FirstName, string LastName);
The example that would have made the most sense would have been to inherit from Record2 in order to show more clearly that the methods/properties/etc will be inherited too
Your line isn't the same as mine - Record1 has a constructor and we need to pass data into that constructor to satisfy it. So, when we inherit from Record1, we also tell it which two properties to send into Record1's constructor.
Thanks a lot for this amazing video! You should definitely get used to the "Ctrl+D" shortcut tho :D
Glad it helped!
Hello Tim, I listened to the whole thing. Very informative, but one last bit you said around 1:13:15 tripped me. You said the record clone is making a shallow copy. But how so? We didn't have any reference types as any properties in the record to begin with. What exactly is shallow copied when you clone a record? Did you mean the reference to the record itself is reused on a clone?
In C# strings are reference types, so in this case when he is creating a new record from a copy of the original and replacing the FirstName value the new instance re-uses the LastName string. This works nicely with multiple levels of records since each instance is immutable so keeping references all over the place as you change only parts of a sub-record is a great performance gain without worrying about state changing underneath you.
45:26 Actually, C# will handle way better that we will, TBH.
I like the idea of having immutable data philosophy and being closer and closer to functional paradigms in this C# new era.
In my opinion, record types comes with lots of extra stuff which is only shadowing what they should be, which is *immutable data structures* . The idea of redeclaring properties and the ability to declare methods makes records less responsable.
*Data structures must store data* , nothing more, nothing else.
If any system needs to hide some properties from a certain data structure or maybe remapping some of their properties, then the system must do it by itself (by implementing filters, mappers, handlers, etc), not the actual data structure.
Oh my god, this video contains huge amount of infomation. Thanks a million!
Glad it was helpful!
5:58 - Records act like value types, but they are not
7:44 - Meaning of "init" keyword
8:03 - Records are immutable
17:22 - Records equality
21:04 - Reference equality
23:27 - Equals() and == are not the same
28:19 - Record deconstructor
32:00 - Copying and modifying records
Great explanation of records in C# 9!
Thanks Tim
You are welcome.
Great video! Thoroughly enjoyable. I would just like to mention a couple of more use cases that records may be useful for. One thing I especially like about them, is how they can be declared with very little additional boiler plate code. They can be declared almost completely in domain language and they are compact, making them easy to read for non-tech oriented domain experts. They can be used as a tool for defining the domain knowledge in code, and being understandable between domain experts and developers! Another thing they can be really useful for is when combined with the new pattern matching features (Especially switch expressions) - They can be used as a reasonable substitute for discriminated unions, as they are able to be derived from - This, combined with pattern matching is a very powerful combination, and I think would probably warrant its own video. I loved the explanations, and I dont know why MS made the choice to allow mutability.
Thanks for sharing
Great video, many thanks Tim! I'm definitely going to recommend your videos to my co-workers.
Please do!
26:48 very dangerous advice here: do not expect 2 objects producing the same hashcode to be equal, that's not a guarantee of the GetHashCode() contract
I'm using records with Dapper now, and it's working out just fine.
Then, I have a different approach to data that needs to be changed/updated: I use the copy feature of records. That makes it distinctly clear exactly which fields will be changed. That's a lot more readable than using classes and having to copy every property. Using a record copy exposes clearly what my intentions are, and leaves much less space for bugs to hide in.
All my records are still immutable, my code is brief and focused on what I do. it's good :)
Really great explanation!
Thanks!
Thank you very much TimCorey you explain very well.
I guess we also can use records for const variables
They won't provide you the same benefits as a const, but they are read-only (mostly).
Huge addition to C#, introduced with a great video!
Thank you!
You're very welcome!
For 40:00 , isn't it due to internal properties are designed to not return through public ToString()?
Yes. I think so. ToString() called from another aasembly on that record type should not expose internals.
Yep, it was because it was protecting me from potential problems (amazing how smart that is).
@@IAmTimCorey is there a way to capture all of these after video creation knowledge into the description? I knew someone would have an answer to this but I had to skim past several comments to get an answer.
Woah as you said new shiny stuff, but "why"... I liked that you answered that question. Great Video
Glad you liked it!
Correct me if I'm wrong here: would a MVC ViewModel be a good place to use Records?
As the Model or the ViewModel? If the model then it could be as long as you don't manipulate the data coming in or going out. It would not really work as the ViewModel because the VM inherits from a class and it allows for data change.
@@IAmTimCorey my bad, yes I did mean the @model declaration at the top of the page. Secondary lesson of the day View's Model is not a ViewModel.
Very informative video. Other resources and even books don't cover that much about records!
10:45 Going one step further, you can use "control dot" to create the property or field from the constructor too, instead of creating them by hand.
Thanks for the Tip
Great Tutorial. I enjoyed it and got a lot out of it.
Thanks for following Tim
Very informative. I did a couple of projects using Akka.Net. Since the objects passed around by Akka must be immutable, records are an excellent way to encapsulate the Akka message data and you don't have to worry about accidentally changing your object.
Thanks for sharing
☝🏾🔥👍🏽
That is a very good deep dive into records.
Thanks!
Wow, I missed the ability to omit a type during instantiation in c# 9. That's nice, my eyes crossed when you first used it.
Good stuff!
Timestamp please?
I certainly found the use case for records in our project. For serverside caching, we calculate the SHA256 hashcode of the requests to identify if the request getting repeated, I see records as a potential replacement for classes there as we don't need to change the incoming requests and want the same hashcode every time the same request comes. Scope to get rid of SHA256.
Nice!
excellent work in this video, thanks
You are welcome.
Dumb question: how do you enable the parameter name shadow in functions calls? Like at 12:40 line 9 , the FirstName and LastName shadows?
You can enable it from : Tools => Options => Text Editor => c# => Advanced => Display inline parameter name hints (experimental)
Mahmoud is correct. You can also watch my previous video to this one (5 features of C# 9), where I show off how to enable this.
Thanks all of you, I was searching for this this.
I would add that records are useful when writing more functional code (c# seems to be migrating to be a hybrid language) and if you are using records it should be records and immutability all the way down.
A developer can run into issues if a record has a class that is mutable making the record shallow immutable. They do a solid 85% of the work for us but you can still shot yourself in the foot.
It is possible. Thanks for sharing.
After watching I was dying to check this out:
static void Main()
{
var r1 = new Record1();
var r2 = new Record1();
Debug.Print($"r1: {r1}");
Debug.Print($"r2: {r2}");
Debug.Print($"r1 == r2: {r1 == r2}");
}
public record Record1()
{
public int RandomNumber => new Random().Next(0, 100);
}
What do you think the last Debug.Print will produce? Take your guess! :D
P.S. Thank you Tim for another great video!
Thanks for all the content you always deliver. I made huge progress in C# over the last year and your videos played a big part on that :)
I have two examples where I'm still not sure what I would use:
1. A Coordinate. Usually it was a prime example for a struct. But with all the features of a record, i would tend to create it as a record now. The easy comparison, an easy destructor, immutability and also the ability to have for example a 3DCoordinate inherit from Coordinate. But this makes me wonder, what are cases to use a struct then? (I almost never used structs in the past, do i don't know that much about those)
2. Let's say we have a simple n:1 relationship between city and person (inhabitants) and none of the cities properties will ever change. Just the objects in the Collection for inhabitants. Would it make sense to use a record here, if i want to guarantee immutability?
Very nice explanation, thanks
You are welcome
Thanks coach!
Any time!
Wonderful explanation, very useful, thank you so much!
Glad it was helpful!
@@IAmTimCorey I've learned a lot from your videos, you really are a great teacher. Greetings from Italy!
10:47 this also works vice versa, when you've defined your members like FirstName and LastName already, .-space on an member and you should get the option to create an constructor with parameters. I believe you can even choose which parameters. While ctor only gives blank constructor, this gives you parameterized constructors.
Nice!
Thank you so much for sharing your vast knowledge with us, Mr. Corey! I think I am going to use this records for my DTO objects. This should work fine I guess because I don’t think they (DTOs) should be mutable.
You are welcome.
Thank you so much for the awesome video! Everything was very well explained! Only thing I was still wondering was if Records can inherit Interfaces or not. I searched the web and I see that they can! :)
Yep, records are classes so they can implement interfaces or inherit from other records.
"Personality, I wish that classes would do this".
Same. For a lot of my classes I simply override ToString() to output a json representation of the class. For the className I simply declare it as a string.
Maybe someday.
Great video. Thanks a lot!
You are welcome.
Records can be use as DTOs?
Yep.
In minute 52:00 what happens if it doesn't inherent from Record1? Wouldn't we still get the same output and options?
Yes, but then User1 would not be a Record1, so we could not assign a User1 object to a Record1 object (not a big deal, just something to note). If we had inherited from Record2, there would be more implications because of the custom code in Record2 that we would be inheriting.
@IamTimCorey
21:09
Wouldn't the values be read from memory and not from disk?
Yeah, slip of the tongue (multiple times).
@@IAmTimCorey I figured, thanks for the clarification.
I might give records a go for a project at work. but I am running into issues with deserializing from JSON into a record with the JsonPropertyName annotation. Are records suppose to work with annotations? I don't get error messages, but my record comes up with no values when i print it, if i take the annotations off it works, but i have to match the Json names exactly, which causes a problem property name casing.
@IAmTimCorey
I figured out doing something like this worked out
public record PlanetRecord(string Name ... rest of the properties)
{
[JsonPropertyName("name")]
public string Name { get; init; } = Name;
... rest of the properties
}
Thank you for this awesome C# Record.
You are welcome.
Hey Tim, love this video. Totally unrelated but I wanted to let you know that if you do ctrl + enter at the end of a line, it will copy it and put in the next line. I saw you doing a manual copy paste :) Nice shortcut when you want to copy and paste a line. Thanks again for the vid.
Great tip!
Nicely done. Thank You!
Thanks for watching!
awesome, thanks again
You are welcome.
I really thank you for this video tutorial
Glad it was helpful!
Hi, I was looking for a spread operator in C#, I think record just fit with the Blazor-state. Great tutorials, thanks.
You are welcome.
thanks Tim for the time and effort , try using CTR+D to duplicate lines in vs
Cool! I didn’t know that.
I have vsVim installed and I do yy p
🥸
Good tip!
Great as always, thank you Mr. Corey! So records are immutable unless you add setters and make them mutable, got it :-).
Thanks for watching!
Wow, its a real deep dive in to record.. I wondering what is a way to use a record in foreach or for loop?
Sure, it works just like any class.
Microsoft should hire you Tim 😉
I'm enjoying being independent.
thank you 👍👌🏻
You are welcome.
I know this is old but once you override the properties you have to make a constructor and deconstructor for the overridden value for it to retain its original functionality, the reason is you changed it to a non-positional record, similar to how a list works. You told the record I need this to behave differently so it pulled the value from its position does the changes you declare then pops it back at the end as it moves in order. So in your example Last name moves to the front of the order and the changed Firstname moves to a new position behind it.
The reason you can use { Get; Set; } is because you need to be able to use them for a public record struct as a struct cant be {get; init; } your case in Do not do is an exploitation of this behavior by hybridizing a loophole. When making the record a struct it takes on the behavior of a value type not a reference type removing the need of immutable.
In fact as an example:
public record struct Point(float x, float y) //initializes as
public struct Point
{
public float X { get; set; }
public float Y { get; set; }
public Point(float x, float y)
{
X = x; Y = y;
}
}
unlike public record Point(float x, float y) //initializes as
public class Point
{
public float X { get; init; }
public float Y { get; init; }
public Point(float x, float y)
{
X = x; Y = y;
}
}
Thanks you for that tutorial. Is it amazing !! :)
Thank you for continuing your learning with Tim.
Many thanks for this video.
I have a couple of queries:
1. Can we place records in a list?
2. Can we use them in as Models in MVVM (E.G. PersonModel)?
Yes, and yes.
@@phizc Thank you.
Vemund is correct, although typically in MVVM you want to make changes to your model. In that case, you would want to use a class.
@@IAmTimCorey Many thanks. That confirms what I thought.
I think records are good candidate for dictionary keys
Great stuff!
Thanks!
@
IAmTimCorey
I like using record's when working with Json (System.Text.Json). I write a record that describe my format, then I serialize and desterilize it when needed. Also, I like using the custom Deconstruct void in my record to do Pattern Matching.(Fun Stuff)
Great!
Great video as always!
just got around to learning this.
1 quick question - where would you define those records?
in classes, we usually define each class in its own file
is this the same for records?
It depends. I like them to have their own file, but sometimes it doesn't make sense.
Hello, Tim. Thank you for this lesson. It was very interesting and useful. But I have a question. As I know all types for collections are reference types. And I can't use List or Array type for property in record because their hash codes will be different for different objects. So how can I work with record type that has to contain collection?
Did you get an answer?
@@guiorgy, Nope. I use records just for value and primitive types. In other cases I use classes to avoid misunderstanding of object immutability.
@@guiorgy, or you can write your own collection type and override GetHashCode and Equals methods. Also there are some types that implement IReadonlyCollection, but I didn't investigate how they work. May be these types can garantee that hash codes will be the same for same collections. Need to check it.
@@levkurbanov431 From a quick Google, seems that readonly collection also don't work. You'd have to extend collections with value semantic Equals and hashes, and if not careful, something else might break, like Json deserialization. In general, reference types without value type semantics should be avoided in Records, I guess.
Source: stackoverflow /a/69840868
Great! 😊
Thanks!
the equality checks make records compelling for me. I worked on a project where we kept multiple versions of documents in mongo, so on ingress (creation) we would find an existing, and compare so as to create a new version. Needless to say, we had to override gethashcode based on each of the properties, and then write/maintain equality tests for each such class.
That sounds like a good option.
Is there a way to force in a project, that all records wil be fully immutable? I meant, lke not allowing having mutable fields in the record...
You can do that with Analyzers. I don't know if anyone's made one yet, but it wouldn't be hard.
It is not baked-in yet. I would like that feature as well.
@@IAmTimCorey would you volunteer submitting such a feature request?
@8:00 you are comparing init properties to the record and that your previous video etc. and that you can set those values via curly braces during construction. Is there a way to do that with Records? When I am trying it, I am required to use the constructor. Is there some option I missed in your video to allow me to use curly braces construction?
If you have a constructor, you need to use that unless you also create an empty constructor (just like with a class).
As someone new to C#, but experienced in a number of other frameworks, paradigms, and languages. You spoke in the use cases about the use of these immutable types (Records) when Parallel processing or reference data is being loaded. I can see how you can hard code a Record, but when you have a Data Source (such as the output of an API in JSON or a set of "Objects" from an ORM to your Database via .NET), I can see how I can easily deal with the ingestion of those into a Class, but can Records be the subject of an ORM easily or do I have to unravel things from a class that duplicates the content that I'm interested in -- inheritance from Class to Record would seem to be a logical way to bridge this. At the end you use composition with the Class having a Record, can that be the subject of an Object Relational Mapping without a ton of manual assignments?
Awesome quality and detailed dive into Records, I'll be on the look out for other subjects on TH-cam and take a look at the courses you mentioned.
I may have typed too soon. You mentioned "Dapper" in your Don't section and EF is not working yet. You see I would like to have an Add/Change/Delete C# App to maintain my massive DataSet and then have Records in my Analysis App with a way to share definition of the Database Access to help with changes between the two. Anyhow, any insight you can impart would get great or another Video would be even better :^)
Well, my video introducing Dapper will help you: th-cam.com/video/Et2khGnrIqc/w-d-xo.html (I also have an Advanced Dapper video if needed). I'll be showing off Records more soon (probably in next week's video, for example).
@@IAmTimCorey Much appreciated!
Thanks Tim! Would be great if you could condense these to 20 min max. Trying to watch this using 2x but it still feels slow.
That is what his 10 minutes series is intended to address - th-cam.com/video/yIXqn1L7p_4/w-d-xo.html