Master the Design of Functional Types in C#
ฝัง
- เผยแพร่เมื่อ 29 เม.ย. 2024
- Check out Dometrain and use code ZORAN for 15% off any course ► dometrain.com/?coupon_code=ZORAN
The 1980s are over. The new C# does not work well with old habits. Get rid of them and dive into modern-day programming.
The road to learning begins with understanding why C# becomes progressively more functional with every new version. Yet, most C# programmers do not know what is functional there or how to use it correctly.
This video explains the fundamental techniques of designing functional types in C#. Learn how to utilize the rich syntactic support for functional types in the modern C# and benefit from the novel design methods.
A word of caution: Anyone attempting to benefit in an object-oriented way from the new functional C# syntax will only find frustration and misery. Open your mind and embark on this journey!
Download source code ► / zoranhorvat
Join Discord server with topics on C# ► codinghelmet.com/go/discord
Enroll course Beginning Object-Oriented Programming with C# ► codinghelmet.com/go/beginning...
Subscribe ► / @zoran-horvat
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
👨 About Me 👨
Hi, I’m Zoran, I have more than 20 years of experience as a software developer, architect, team lead, and more. I have been programming in C# since its inception in the early 2000s. Since 2017 I have started publishing professional video courses at Pluralsight and Udemy and by this point, there are over 100 hours of the highest-quality videos you can watch on those platforms. On my TH-cam channel, you can find shorter video forms focused on clarifying practical issues in coding, design, and architecture of .NET applications.❤️
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
⚡️COPYRIGHT NOTICE:
The Copyright Laws of the United States recognize a “fair use” of copyrighted content. Section 107 of the U.S. Copyright Act states: “Notwithstanding the provisions of sections 106 and 106A, the fair use of a copyrighted work, including such use by reproduction in copies or phono records or by any other means specified by that section, for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright." This video and our youtube channel, in general, may contain certain copyrighted works that were not specifically authorized to be used by the copyright holder(s), but which we believe in good faith are protected by federal law and the Fair use doctrine for one or more of the reasons noted above.
#csharp #dotnet #functionalprogramming - วิทยาศาสตร์และเทคโนโลยี
Check out Dometrain and use code ZORAN for 15% off any course ► dometrain.com/?coupon_code=ZORAN
Download source code ► www.patreon.com/zoranhorvat
Join Discord server with topics on C# ► codinghelmet.com/go/discord
Enroll course *Beginning Object-Oriented Programming with C#* ► codinghelmet.com/go/beginning-oop-with-csharp
Subscribe ► th-cam.com/channels/xsWfh8LCcn55mFB6zGBT1g.html
Just give me my dang tagged unions already
I'm with you!
One day 🙏
@@zoran-horvat What is tagged unions ? I know about discriminated unions in F# and C#, are these concepts "tagged unions" and "discriminated unions" the same ?
@@tore28 It is a union which also contains a "tag" field (usually just a byte would suffice). The tag identifies the type of the content stored in the rest of the object. Tagged unions are used to support type switching on polymorphic types that do not possess a header common in object-oriented classes.
Thanks for the video. Especially the naming for types helped me since I encountered the feeling that when I code in functional way, my functions seem "not in right spot" or "not highlighted enough". Functions are in this case more important to me since they hold logic while types simply hold data. Thank you very much.
Task is the most prominent Monad in c#, right?
Right. The other most prominent one is IEnumerable.
guys stoooop
So many ideas fell into place while watching this. thank you.
In F# I regularly have same name for type and module: type Book, module Book, because F# does not have restriction that C# has.
Yes, in C# both are just classes.
Actively trying to learn pure functional approach by learning Haskell, this video is quite useful for me, I can see an example of this style in familiar language to reach more understanding. Thank you.
Cool video.
Functional concepts like bind are nice.
IMO things like ErrorOr are the simplest way to implement sane functional programming in C#. Procedural functional programming if you will
Async/await is basically a monad 🤷♂️
Thank you, my understanding of the difference between OOP and functional was way off before your video.
Excellent video, you are top-tier.
If you had to model different types of xml nodes with different rules and linq XElement is not enough, would you go the functional route?
Most nodes types would be known and not change in the future.
And what about text inside those nodes?
Im thinking about modelling to the smallest bit, like an object Character that has an index, a start column and end column (imagine that you can place the cursor before or after).
I think Functional is orthogonal to your question! If you need a custom type for a node, that's fine. If you want to model the behavior in the class, then go OOP. If you want to model the behavior in extensions, separate from the state - go functional!
I like the Option monad, but I absolutely love the Result monad (because it can provide information about the outcomes of operations).
Hey....perhaps you can do a vid on defining extension methods of `Select` and `SelectMany` so that one can use your Option type with the so-called 'query syntax'??
Option monad is great.
The best implementation IMO is ErrorOr. It's not Result in the strictest sense, but it gives a great failure path, and is really easy for most developers to understand
Thanks for the video
Very great explanation
Do you have a SOLID or Design Pattern course?
I have several courses on design patterns on Pluralsight and Udemy. Not on SOLID.
@@zoran-horvat
I'm so excited to watch it all. I'll do it
Thank you very much
Hi. Do you have a video or example code about the Either-Monad?
I will get to that soon. In the meantime, I have an explanation of Either in my Pluralsight course on functional programming with C#.
I don't understand, the variants are internal but they can be freely accessed in the same project, may be that was kind of and abstract example?
That is only true as long as they are in the same assembly. If I moved the Models namespace into a separate assembly, which I would certainly do as the model grows large, then all code that references the internal members would fail to build.
Thanks Zoran! I guess SelectMany could be one of the monads you're referring to?
More likely the IEnumerable... Deferred execution (or lazy enumeration) of LINQ + IEnumerable is how monads also work. You declare the HOW without actually performing any operation, yet; only declarations. Once all set, you can unwrap the monad (or enumerate the enumerable), where each operation declared will be now performed.
@@keenkidash7616 I don't believe deferred execution is necessary for a monad definition. It's just a nice feature (or footgun) that functional libraries tend to have.
@@Robert-yw5ms A monad requires a factory which elevates a regular value into a monad, and the Bind function, plus a few other rules known as the monad laws.
I am ok with FP where it makes sense. But that Printable example is not such a case.
It would make much more sense and avoid refactoring many places if the NameType simply defined a contract which concrete implementations must adhere to.
Whenever a developer needs a new NameType, they add a new class, implement the contract, and the rest of the code keeps working without changes.
Overall I am learning a lot from your videos but to me this example in particular is where OOP wins against FP - and that's ok, each has their advantages. As always, use the right tool for the job.
While I appreciate how you've gone about it, I find the Book vs BookType confusing. If all Book is doing is providing static creation methods, why does it have to be a separate class (and thus the segue into how to name it), rather than just part of the record class? Is it to avoid the temptation of directly calling new Book()?
The reason is that it will not be all the book does. The next step from here is to start defining behavior, and we do that in the form of functions defined separately from the data type.
The greatest benefit from functional programming is that, unlike in OOP, different portions of behavior are defined in different packages.
@@zoran-horvat But it is perfectly fine to declare static functions in the (abstract) super type. Thus, the naming theme "***Type" becomes superfluous.
@@thomasschroter3802 Yes, but that goes against the general idea that those functions can also be defined in another module or namespace. Why make a special cade of some functions, but treat others regularly?
I always have a hard time knowing when to use this. When using c# why not keep using OOP. Why do i go functional, and when do i use it?
When you work on data with many (100 or more) rules. In parts of web backends not uncommon for example. You can do it with with oop but it is easier to read in functional stile.
Any model where the data model is known and never (or rarely) changes, but behavior grows every day, will likely be a good candidate for functional modeling. Doing the same work in an OO model would require several times longer code and will cause enormous issues.
Funny enough, that is a common situation in business operations and connected services today.
Use both
Use extension classes to create functionality.
Use objects/classes for data.
This mental model is easy to understand: data goes into a function, and comes out in a different shape
Love functional programming and Zoran is guru in it. Who would like to dive deeply I recommend listening his Functional Programming lectures on Pluralsight
Is it Nullable? :)
No, it is a regular reference type that can be null. The Nullable generic struct is only used on value types.
CreateMany() code in 10:27 seems very strict, I will prefer it like if All are null then return null, else return only the one are not null. so if authors are ["a", "b", null, "c"] => will return ["a", "b", "c"] instead of null.
If you considered the idea of a smart constructor, you would see that it doesn't make much sense to request _all_ contained objects to fail before failing.
What you are describing is the Bind function on IEnumerable. It is useful, but in different conditions.
Why did you check string name for null but haven't done string? firstName in the function parameterz. Just forgot '?' ? (My ide would say it's always false)
That was just a line to show the exception in action and then to delete it.
However, if the factory function was supposed to be called from the outside, then non-nullable references can still be set to null and you would have to guard.
I don't see the benefit of functional programing. Where do you put the breakpoints in those shiny onliner flatMap functions? Also, besidedes saving vertical screen space, what other benefits do I get from fucntiononal programming.. (Hard mode: no use of "mutability", "pure", "side-effects" words)
What breakpoints?
Do you really develop software using breakpoints?
@@zoran-horvat unit tests, and breakpoints.. nothing against some functional programming stuff, like pattern matching. but going all in, seems a bit overkill. specially since my brain is used to oop, using guards, throwning exception for early exit. basically I bought the Uncle Bob memes.
btw, are you from croatia? zakon da sunarodnjak radi tako dobre tutoriale, svaka cast.
@@DavidSmith-ef4eh Zapravo sam iz Beograda, mada imam rodbinu u Vinkovcima i okolini a prezime vučem iz Slavonske Požege 18. vek otprilike.
@@zoran-horvat svejedno, svaka cast na dobrom kanalu. jedan od boljih opcenito, tek danas slucajno naletio na to... ali vidi se kvaliteta i sadrzaja i produkcije.
I think LINQ SelectMany is kind of monad?
IEnumerable with SelectMany and a creation function (e g. ToList) is the monad.
@@zoran-horvat I don't see what ToList has to do with IEnumerable being a monad. You say 'creation function' so I think you must be talking about the unit needed for a monad (since that's how you create a monad from a value), but ToList doesn't serve as a unit in any sense.
@@Robert-yw5ms It does serve as Unit, because it supports wrapping a single object into an object implementing IEnumerable. Other example is the array literal, which is also constructing IEnumerable.
@@Robert-yw5ms Aaargh you are right - I wanted to say new List, which supports the initializer syntax, not ToList.
I wrote 100 comments today, please understand me :)
Linq library is Monadic
It is, indeed.
Overall I think the state of functional programming in java to be superior to c#. Especially 14:20 java supports closed hierarchies enabling exhaustive type pattern matching.
I also think it's overcomplicated to hide the varients and create a match function. The sum type NameType with its varients FullNameType and MononymType can do these things with 1st class c# features (switch expressions) in the user code. Also separating the static code from the type is unnecessary and confusing. Just put the factory functions inside the record. This way you save yourself from the naming dilemma as well.
C# switch expressions are still cumbersome with no exhaustive checking. They must always include a throw instruction, which makes every screen if code look ugly. Once we get that feature, I guess I will stop favoring the Match method.
You can never guarantee that no one will ever extend your abstract class. It can be done in other assemblies. Sometimes even if the type is internal. If you skip the default switch case C# will add it for you. Try and test.
@@rafazieba9982 I am talking about the fact that java has a compile time and runtime mechanism to prevent unallowed inheritance. The syntax is somewhat confusing from a c# perspective as it uses the sealed keyword which means something different in c# and java.
@@redcrafterlppa303 I don't know of any Java syntax that allows those two classes to inherit but nothing else.
@@rafazieba9982
public sealed class Super permits Sub1, Sub2 {}
final class Sub1 extends Super {}
final class Sub2 extends Super {}
Super foo = new Sub2();
switch (foo) {
case Sub1 s1 -> {}
case Sub2 s2 -> {}
//no default (exhaustive)
}
If one feels like writing C# functionally to tee, isn't it better to just go with F#? Golden hammer..? 🤔
Not necessarily, because C# allows you to write part of the domain functionally and the rest object-oriented. It also supports rich procedural syntax, which is useful in performance-critical portions of code.
There are many benefits from using C# in design.
@zoran-horvat
F# allows you to write terse OOP code just like C#. And allows you to write procedural code.
Of course there are some hurdles, especially when you need to cast to interface due to Hindley-Milner type inference in F#, but so many hurdles for FP design exist in C# - primarily OOP oriented, no inbuilt mutability, non-exhaustive pattern matching, lack of native discriminated unions (and standard Optional objects, etc.)
Given that .NET is polyglot environment, why not use F# for functional design and C# for OOP?
@@dusanknezevic9072 Then there is the inverse question: why not use C#?
You cannot mix languages within the same project, so .NET will not help you with that.
consuming c# libraries, whose are a lot of, from within f# sometimes is hard and not worth the effort
@@zoran-horvatnot within the same project yeah but you can have two projects one for each that can work interchangeably (you do have to rebuild for the other project to update but that’s the only issue I know of) however you do need to make sure not to use F# types for stuff the C# project will use or it will get messy but while it can be done I would also not recommend doing so. There’s features in both F# and C# I wish where in the other that would make both a good option for this
doesnt that mean if someone creates a nametype the printable doesnt work anymore
I'm not sure I understand the question.
Yes, this means if someone creates a new type of name and attempts to use it, there's some broken down-stream effect because the logic doesn't know how to process it. Creating a new type will mean tracking down all strategy patterns that switch on the type and updating them. People will say "just have unit tests!," but I think that's insufficient, it makes me kind of dislike this approach unless used very sparingly.
As Zoran said, C# doesn't have the tools to do exhaustive pattern matching. There are a number of libraries doing it. The goal is that if you add a new type, then the projects using pattern matching on that type will not compile, and the error will tell you that you are missing a type.
@@andrewshirley9240 There is a great misconception about functional types where many programmers believe they are somehow less powerful than OOP classes. The truth is that those two concepts are orthogonal.
While all FP consumers would fail to compile when a new type variant is added, the same stands for subclasses when the new method is added to the base class.
The reason why FP is so practical today is that business applications require frequent addition of functions, but not new variants of existing types. That is precisely what hurts in object-oriented design.
ahhh, just use F#!
Nope.
when a new author type is added, the formatting function will need to be redone (add a new Func parameter). If there are too many inherited types, your code will turn into noodles.
Now listen to this - it is your words:
When a new method is added to the base class, all subclasses will need to be redone (add a new method override). If there are too many inherited methods, your code will turn into noodles.
It is genuinely funny how easy it comes to criticize functional design, but fail to see that it is in every respect a mirror image of the object-oriented design. Every strength of one is the weakness of the other, but also - vice versa.
I hope you will give this detail a fair portion of thinking before making another comment.
@@zoran-horvat Just curious how would you handle the match function when you get new NameTypes? f.e. We need to add support for a middle name, maiden name and a pseudonym(alias)
@@mathys-dikmans Adding new subtypes breaks all functions in functional programming the same way as adding a new method breaks all subclasses in object-oriented programming.
The reason why functional programming is gaining so much popularity in recent years lies in understanding that we add new functions much more frequently than new variants in discriminated unions. Therefore, while the problem you are referring to does exist, it is significantly less frequent in practice than the corresponding problem in the design of classes.