This course is archived!

Login to bookmark this video

The Patterns Behind Doctrine

Share this awesome video!



SymfonyCon 2018 Presentation by Denis Brumann

How does Doctrine talk to your database? What are Data Mapper, Unit Of Work, and Identity Map? These are the questions I want to answer in this talk. We will look at how Doctrine ORM implements them and what they are there for. Finally we will look at how they compare to Active Record and what the benefits and drawbacks are to help you choose which one fits your needs best.

A lot of things, um, I hope it's not too fast. But in any case I will still be around afterwards, so if you want to talk about any of those topics I will be talking about, we can talk about that later on in more detail. So first of all, because I already know that in this room it's really hard to see the slides if you're in the back of the room. The qr code goes to to the talk. I already uploaded the slides with some additional pages in there that you just have to skip over. And yeah, you can just go to the talk/c1b8c is the talk id and, and you can just get the slides and you watch them on speaker deck as well if you don't want to follow on the screen or things are hard to read.

Hey Denis!

So, um, with that out of the way. Um, hello again, my name is Denis Brumann. I work for SensioLabs in Germany as a software developer in Berlin. And you can reach me via email denis.brumann [at] and also on twitter and GitHub with my initial and last name.

Disclaimer: Doctrine 2 Vs Doctrine 3

And before I start I want to make clear that everything I will talk about is about Doctrine 2 not Doctrine 3. It's not out yet, but it's already under active development. The master branch already points to Doctrine 3. Things might still apply in some way or another, but, but it will probably not be the same. For example, the unit of work is probably not stick around. Um, so keep in mind this is how things are right now, but it's still helpful to know because if you want to do this switch to Doctrine 3, you probably want to know how Doctrine 2 works under the hood to basically know does this change affect me that they introduced. And do I have to look at anything that maybe could cause a bug in my code.

Introducing The Domain: Order & OrderItem

So we will start by just looking at a simplified domain that we basically we'll follow through the, through the whole talk. And I only have two entities in the domain. So it's kind of easy. It's probably easier than all the entities setups that you have in your projects, but it will be enough to to look at all these things. So first of all, we have an Order that just has an id. It's an auto-increment id, very basic stuff that you will probably all know. We have a timestamp from when that order was created and then it's basically just a list of OtherItems which we'll look at later. And everything is set via getters and setters just like most people are familiar with. There are other nicer ways of dealing with things, with like rich domain models, but where it's more expressive, but we will just look at something that's kind of ordinary.

And the OrderItem is the same. It has the same auto-incrementing id. It's linked back to the Order. Um, that's a necessity because OneToMany requires ManyToOne. We will look at that. It has a name for, for the item that we want to store in our Order. And it also has a price which is a string, which might seem weird at first. But if you ever worked in an e-commerce system using floats, it's really a pain. It just causes rounding errors and, and, and it's just horrible to work with. So one workaround is to use ints, so integers. So basically instead of using the euro amount, use the cent amount: 100 cents are one euro and that's it. And that makes things a little bit safer. And I pushed it to the next level and used string and you will see later on why we do this.

So let's look at the code. So you probably all have seen this. It's a very basic entity. We have the usual annotations. In this case we also add a custom repository class. You probably have seen that as well. We give our table a specific name. For the OrderItem it doesn't matter as much. For the Order, it's more important because order is a reserved word in most dbms. So that kind of messes up things. But when you work with Doctrine, um, so it's nice to, to give it an alternative name. As I already said, the auto-increment integer, that's pretty much what you all know. Then we link to our Order. Um, the interesting thing is that we basically tell the JoinColumn that this OrderItem always needs to be attached to an Order because it doesn't really make much sense to have an OrderItem just loosely hanging around in the database. So we just ensure this with the JoinColumn(nullable=false). And the rest is just basic stuff. We get things, we set things. You probably all have seen this, you can auto-generate it. So it's really simple.

With Order it's kind of the same thing. We have an auto-incrementing id, we have the relationship this time a @OneToMany that maps back to the OrderItem $order property. And if you think about the, the table layout, it kind of gets clear why we needed the other relationship from the OrderItem back to the Order because obviously the, the order ID is stored inside the OrderItem. It links back to that. So if we were just to have this relationship with, on our Order, that that wouldn't be a reference to: okay, where is information stored? It's in this different table. So the, the OneToMany relationship is the only one where you actually have to have a bi-directional association. In other cases you can just have a unidirectional association.

And the other important thing here is the cascade={"persist"}, which come, will come in later, but just keep it in the back of your head that this is important. And yeah, for the timestamp I just use the DateTimeImmutable object. It's one of the types that Doctrine supports. We initialize the items that we want to put into our Order is an ArrayCollection. You probably have seen it 100 times and more. Um, in my case, I return the array instead of the ArrayCollection. This is just something I like, but there's no specific reason to do that. Um, and it doesn't really come up later, so I just want to point it out that this is a quirk I have, you don't have to do that. Um, what is important is the addItem(), we basically, because we have a bidirectional association, we have to make sure that both the Order and the OrderItem know what object they are linked to. So in this case, I set the Order um, to $this on the $item and also add this item to our current Order if that makes sense.

Um, and yeah, this is where the string part for the price comes in. I use the BC math extension for PHP. If you use modern php where you have strict-typing and everything, you can't just pass in an integer or float, it has to be a string, otherwise you will get a type error and that's why I use the string all the way through and it will just add those strings, strings up from the items and then you get the final total price. So it's really simple math. It's just doing plus current total plus the next item.


Okay. So those are the basics. Um, I know that was kind of fast, but I, I kind of hope that you got the structure in your head, you know, what we're talking about in terms of what kind of objects we're dealing with because this is all we will use throughout.

And yeah, just assume that with our very simplified e-commerce system that we have now, we want to do a checkout. We select items from a page - I don't have everything built for this case - it's just, let's assume it happened. And then we pass things to a Checkout service. I don't use the service suffix here, but this is just a basic service class. Um, I pass in the EntityManager because we want to store the Order when we do a checkout with all the OrderItems. And this is what the actual checkout method looks like. In my case I just get some $cartItems which are basically products that, uh, in a DTO entity or whatever that is used on the frontend for displaying my product items in the cart and checkout. And now I want to transform them to, to an OrderItem.

It's basically to make sure that when the product changes, I change the price, my OrderItem is supposed to still have that price because I will still want to charge the customer the original price they used when they used the checkout. And then I just do some mapping. I map the name and the price from the $cartItem to the OrderItem. I add that OrderItem to my existing Order that I created up there in the top, and then I just save that Order and I flush the entity manager. So you probably have all seen this.

This is really basic stuff and still, there are some really interesting things going on behind the scenes. Because if you think about it, first of all, we are only persisting the Order but the OrderItems will be saved as well. So Doctrine, for some reason knows it has to save those items into that separate table. Um, and it also knows that has to do insert instead of just an update. So let's look at why this happens? How does Doctrine know, which items to save?

Unit of Work

And that is really interesting and it works based on the unit of work. This is a really huge object and, and it's kind of a basic pattern. So, um, the, the example on the right is basically what Martin Fowler has in his pattern catalog. I omitted a few methods that we don't really care about, but basically this is an object that stores our changes for the database and, and keeps them in memory until we flush and we actually want to write them. And then this unit of work, will figure out what writes it has to do: if it's an insert, if it's an update and which order things have to happen. And that's really a lot of stuff going on there and it's a huge class right now, which is probably why in Doctrine 3 that they want to split it up a little bit. Um, and it's also kind of ugly to look into it. Um, so it's fun to dive into it if you want to do it. If you have to do it because things go wrong, then it's just a pain.

So I want to save you from doing that and just enjoy looking into it for the fun of it. Um, and you can already see that there are some states for, for insertions, for updates, for deletions. We have the entityChangeSets, so what changes need to be written to the database. And more importantly we see some of the methods that look a lot of similar to the methods on the EntityManager. Um, so for example, we have a persist() method that looks basically the same. The remove() method, the flush() method that kind of looks like the commit() method. So that, there's some kind of overlap. And that's no, no that's intentional because basically the EntityManager will pass its called to the UnitOfWork and the UnitOfWork will store all those things.

So when we pass in the Order, the UnitOfWork will figure out: ah those other properties that map to a database table, I have to basically save the data that's in there for later change. And I will also have to remember that this object that they wanted to persist, I have to basically write that to the database as a new object based on the state.

And the reason why this thing works with the OtherItems as well as because we have this cascade persist in there. Because when we persist our OrderItem, what will always happen is that the EntityManager will also look through the associations on our entity. So on our Order it will look at the the items property and say, "Oh this is an association, the items in there I probably have to save as well". And then just loops over them and sees either if we persist them manually or if we have the cascade persist, then obviously Doctrine knows everything that I put in there I have to store in the database as well on an insert.

And yeah, this is basically what happens. So we pass in the Order and then Doctrine figures out, the OrderItems in there, I have INSERT them as well and have to write to the changesets. And this is roughly equivalent to adding a second persist in our foreach loop for each item. There are some subtle differences, but basically it works out the same. And if you don't do either of those things, so neither the cascade annotation or persisting all the objects manually, what you will get this error. Because, obviously, Doctrine knows that some objects should be stored in the database because we have the association there, but it doesn't want to write this unless we tell Doctrine to do so. Because it would be kind of bad that when Doctrine thinks:

yeah, I know Dennis is kinda lazy and he probably forgot to do the annotation. I will just save it for him.

And that could be fine, but if it just writes stuff to the database that I really didn't want to have, then yeah, I'm in trouble. So Doctrine says no, I won't touch that. Figure it out yourself. Do you really want to persist this? Then either call persist or cascade or just move it out of the collection. So this is something you have to be aware of when you work with Doctrine that unintentionally non-persistent entities that come up in the changeset, Doctrine will not touch them. You have to figure things out. It will give you a nice error like you saw before where it tells you what to do, but still you have to check if: is this really something that has to go into the database? Probably yes, and then you have to act on it and do that. And also you have to add this cascade annotation to make sure that this happens automatically for you. And obviously only on the entity that is supposed to cascade. So on the OrderItem we didn't cascade for the Order, because I decided that I always want to save the Order with the items but not an item and then create a new Order for that item. Because in my e-commerce context, that doesn't make sense. If you have this context where this makes sense and your item can be the primary object, basically, then you can just do that: just add the cascade and it will also save the new Order through the item as well when you persist one item. And Doctrine will also figure out in what order to do that.


Okay. So much INSERT, there was already like really, really lots of stuff in there. And it's basically getting worse, but at least we have nice pictures. So let's think about like we're inserting new orders, we do our checkouts and, and we now want to have a backend where we want to look at our orders. And we want to have some additional details. So in our order list, we don't just want to show the order number, we want to see how many items are in there, what will be the total price for that. And what you can see on the bottom is that this will issue 2 database queries which might seem odd to some. Um, because yeah, it's just, we just wanted our order list and that could be one query.

And the reason this happens is you don't have to look at the full Twig template, but this is basically the bottom part that that shows how many items are in there and what they cost - the total price. And this is our Order, method, getTotal() price that iterates over the items. So I'm not calling the items themselves, the, the, the method does, but still Doctrine recognizes: ah, those items I need to get somewhere. And it will just issue the second query for those items.

And this can get bad if you have more than one item in your overview or one order in your overview, because for every Order it has to fetch the items. So if we have five orders, it issues one query for getting all the orders and then another query for every order to get its items. This is often referred to as the N+1 problem. So Doctrine will basically issue more queries than you might think it will. And yeah, this is basically what it looks like in a profiler. So you can see that on the bottom you have the query for the order list, fetching all the orders and then for every order of 5 times you get the same SELECT statement, um, for the items fetching all the items for that order.

Lazy Loading

And the reason this happens is lazy loading. You probably have heard that before. And how it works in Doctrine is it uses a so-called proxy pattern. So when you call getOrderItems() or even just accessing the items property on your Order, Doctrine has not loaded this beforehand, because we only wanted the orders in the beginning and our controller. Instead it will see that those items are not loaded, load from the database. And once that is done and will return those other items and it will keep them in there. So it will not do the same query on the same object over and over again. It just remembers: ah these items were not fetched the first time and that's why I'm fetching it now. And then subsequent calls will get those items we previously fetched.

And what's kind of responsible is the, I already mentioned the proxy, the proxy objects that Doctrine generates - at least if you're not in debug mode, so, um, it might create the folder but you will probably not see files on there, so just do debug false and you will see those proxy files and we can look at them. They are huge and ugly and, and I never had to look at them in any real cases. It's just basically to show you what they look like. Um, you can already see that it's just an object over our original Order. That's also why we can't have, entities final, because something has to extend them - the proxy object. And then have an initializer and then you have tons and tons of code that basically is unreadable and, and you don't have to, it's automatically generated and you don't have to care about it. But you can already see that if I want to get those items, some initializer is looked at it, and it will invoke a method. And this initializer is basically responsible for doing the querying in the background. And then it will just get the items as our method on our parent object basically described. So, um, this is the magic behind it, so to speak.

And this is how it looks as UML. I hope I got the arrows right the direction and the arrow tip. I, I'm, I'm not a computer scientist, so I have never learned this and I never get this right. Um, but yeah, this is basically what it looks like. You have a wrapper on your Order object on the right that has the same methods, it just has some additional stuff on there that you don't really have to care about. You don't have to access yourself. Doctrine does this under the hood for you.

Fetch Modes

And this is fine if it's actually what you want. So for example, if you fetch orders and you don't want to access the items, it's really good that doctrine doesn't fetch the items every time you, you get this order. But when you want to have the items and it issues these additional queries, then you kind of want to address this in some cases where it makes the site slow.

And what you, what you can do is add another option to the annotation and say: this relationship I want to fetch eagerly, so always fetch those items. Um, obviously this has the same drawback as fetching things lazily does because now everything is always fetched. And as I said, if we have cases where we just want to work with the Order without items, we now do a query that we don't really need. So this might not work as well, but, but it's a nice and easy way to address this issue to, to not lazy-load.

The other kind of more involved approach is since we have our own repositories and we can write our own DQL queries, we can just tell Doctrine to join the items in certain cases where we define the method and we just call this method. So in this case, findOrderWithItems(). And the interesting bit is the join part. You can see that this is DQ, it works on the object level. So we're not talking to the database table, we're talking about our objects. In this case, the Order object that has alias o, it has some items property - we already know that - and this item's property, everything that's in there, in the association, that's what we want to join. And that's also what we want to select because if you don't select then it will just join it and throw it away when hydrating the object. That's basically if you want to ask something from the items without actually fetching the items, then this can be useful.

And so this is a more involved query because you actually have to write the query other than just doing find(), but, but it can already help you that I can now figure out my cases where I want to do Order with items and Order without items.

Custom Object Hydration

Another approach that I personally like even better is custom object hydration. Um, so take for example, a DTO object that looks like this. It's probably really hard to read. So we have an OrderSummary and it describes all the information we want to see in our Order list. We have the Order id that we want to display, the created timestamp and then the item count and the item total. Um, and then we have just getters, we don't want to change this data because this is basically a read model on our database and we just want to fill this model with the data. And we can still use the DQL query. Let's go through it, like not in chronological order. So first we see that we fetch all the stuff that we need for our OrderSummary and the interesting part here is that we count the items in the database and also we use the SUM method in the database. So we don't have to run the PHP query stuff. We can actually do this in the database which is really fast for doing the kind of integer operations. Then we still fetch the data as if it were the Order object. So we say every data that you would get from the Order object, just just get it. And then we extract the data we want. And then we just create a new object with our DTO in there. And basically it will map the data not to the Order object but to our DTO. And this will also perform only one query and it has all the data in there and we don't even need all the item information that we don't need and we just need the stuff, we get the stuff that we need. And this is what it would look in plain SQL. So probably a little bit simpler, but yeah, the, you just have to translate that back to DQL. Or you can even use sql and use the ResultSetMapping to map it back to the object instead.

And so, just to remind you, this is the queries that we had before. And this is what we get now: we have one query you, can see in the bottom, it kind of looks like the SQL query we just saw, Doctrine just uses different identifiers. And it's also a lot faster, like 3 ms, roughly 3 ms. And 3 ms doesn't sound like a lot, but obviously it adds up. If you have more than six items, like 1,000 orders.

And yeah. So basically your takeaway for this is: Doctrine loads, associated entities lazily by default, which can be good, but it can also be a problem for querying. So you can do custom queries with a join and then select those items yourself. You, you can do a custom hydrations, either into a DTO object or if you just need a scalar value, like a count, you can just get this as well and you don't even have to wrap it into an object. And this can improve read performance tremendously for you.

Identity Map

And something that kind of relates to that, to optimizing querying, is identity map. This is also a pattern that that's wrapped inside the UnitOfWork. So whenever you call the EntityManager find() method with the object class name like Order::class and then the id you want to fetch. What will happen is, the first thing that the EntityManager does is look into the UnitOfWork, into the identity map to see if someone, well actually, if we performed this query earlier before and we already have the object in this identity map. And if we do, then it will just get the object from the identity map and don't even issue a query. If not, then obviously it will just do the database query as before the, the typical find() method and it's fine.

So, um, just to, to basically see how this works. This is very simplified code with just some, some dump in there to see the results. So we do the same find() twice and in between we look at the UnitOfWork and at the end we looked at the $order and the $sameOrder, um, are actually the same object.

And yeah, this is what it looks like. So the identity map contains the class name and then an array where the ID is the key and the object itself is the value. And this is how Doctrine looks it up. Like, okay, I have this class, I have this id, the user wants that, so I have to pass this object on. And on the bottom you see that yea, both objects are actually the same identical object. So every change you do to the object is basically kept and the when you do find again, then you get the object that you already had in memory, it doesn't issue an additional query. Um, and you can see that, on the bottom as well, there was only one database query in this case. And this is basically for our Order object with the find items, it will, at the same time also fetch those items as well and put them into the identity map as well. So if you want to fetch an OrderItem later on by its id, then it will also not issue a query because it already got this item through the Order with the OrderItems. And obviously the downside is that it keeps a lot of data in the memory. But also it's probably doing one request, you won't have like dozens and dozens of entities that it has to keep in the identity map and, and has to work with.

And yeah, so basically that's another important point when you want to improve your read performance, you have to look out that when you use the repository's find() method, it uses the criteria array thing, which is not the same as the EntityManager find() method. So it will actually perform the query twice. It will still fetch the data from the identity map, but it does the query because it can't figure out that, even if you just fetch the id, the find method or your custom query, the EntityManager, the UnitOfWork, don't know how to introspect it and figure it out. But if you're using EntityManager::find(), then you can actually have fewer queries than you would expect.


Okay. So we looked at INSERT, we looked at the find() method. Let's also look at updates. Um, so let's assume we have an object, we create it new. And then we set the ID. Some people might assume that since my object now has an ID, I can just update this object if it's already in the database. And the thing is, obviously, this will not work, because the UnitOfWork keeps track of an entity's state. And since it didn't know the object before, it, it doesn't know what to do. It assumes it's a new object. And so it wants to insert that new object. Because internally, Doctrine keeps track of entities not by the ID but by the object hash. And so Doctrine has a set of states that it kind of looks at and, and tries to sort entities into. Obviously, for new entities, the new state, the managed state, when UnitOfWork knows of this object, it's already managed somehow, that that's both new objects, updated objects and deleted objects and detached objects. So objects we don't, no longer want to maintain via the UnitOfWork and via the EntityManager. So all the changes we do to that object, they will not be persisted, they will not be saved. And removed objects, or something that's basically said that: once you do flush, we want to delete this.

This is how it looks on the code. Um, there are some nice explanations in there. And so basically what happens is, since we created a new object, not via the UnitOfWork and the EntityManager, but by ourselves, the UnitOfWork doesn't know what kind of object is this, and it says: ah, this must be new, I will give it the new state and I will insert it. And on the other hand, if we fetch an object via the EntityManager, the UnitOfWork now knows that this entity was fetched before, it's already part of the UnitOfWork inside the identity map for example. And that's why, when we do this, we no longer need to persist obviously, because it's already in there. Now it knows that I can update this object: this is not something new, this is something I have to update. And so basically, for you, what you have to look out for is, just, kind of, keep in mind that the UnitOfWork manages the state for you. And you don't, you have to make sure that everything you want to insert is part of the inserts sets and so on. So yeah, it's, it's just, it's a lot. I know last session, but we're almost done.

Active Record

So finally, which is kind of unrelated to everything before but, but I promise it in the abstract, so I figured I have to deliver it, is a kind of comparison to, to active record.

So I don't know how many of you know that Doctrine uses the so-called "data mapper" pattern to map things. And another famous approach as active record. Actually Doctrine 1 used it, Propel was another famous object relational mapper, that uses active record pattern. In Ruby on Rails you have the, I don't know what it's called. And obviously Laravel is famously using Eloquent, which is also using the active record pattern. And what the active record pattern does is it wraps an object in a way that everything related to the database is part of this object. So usually what happens is you extend the base model, base record or whatever, and in this base record you have all the database-related stuff. You can query things, you can save things and then you create your new entity, with some restrictions as to how the entity is created. And and then you can do all the database operations on there. So basically the models allow for, for querying the data and you don't need special classes like the EntityManager or repository or anything else, you just do everything for the record.

So this is how it looked in doctrine 1: the abstract Doctrine_Record class. This would be what you would extend. And you can already see that there are some states in there, so it kind of still does all the same things that Doctrine 2 does with the keeping the entity states, for example. So there's this kind of the same feature set, but everything is part of this basic model. And sure there are things that work differently but there's even a proxy thing, so it kind of looks the same. So just to give you kind of look at how you would act on an active record kind of model. And I didn't put the, the entities on there as a record, because based on the library they will look differently anyway.

But roughly, this is what would change. So first of all, since we don't have the EntityManager anymore, all operations are on our model, we just call an $order->save() and this bubbles down to the record and this is where everything is implemented, it uses a connection. Same with an update. Basically instead of getting the EntityManager to find our object, we use a static method called for find() for example or query to get a new Order object. And obviously the flush() has to go as well. We still do the $order->save(). and so we don't need any additional classes, which might seem nice. We actually saw that the code kind of gets a little bit more compact, so that's less stuff to do.

And just just to keep kind of a record of what are the benefits and drawbacks of it. Um, because it, it really depends on what your use cases. Um, so if you look at the data mapper, um, what's really neat is that basically you don't have to care about all these additional classes doing all these things. Everything is inside my model and I don't have to to remember to inject anything. I can just work with my object and do everything on it that I want to do, like saving stuff which can be nice. It's, I don't have to use injection in the service to get the EntityManager, I'll just pass the object and say save. And that's it.

And obviously this comes with a downside that, that my object has to fit some conventions that this record gives me, to, to basically be able to figure out what stuff to write and what not. And with Doctrine you do it very implicit, ah explicitly. You put the annotations on there, you have the yaml file. But with an active record, you basically have to figure out what do they want me to do? Where do I have to store my properties, how do I give it the table name? Um, maybe there are some limitations, like usually you can't use the constructor because this initializes some, some basic database stuff which can be annoying, when you want to do like, rich domain models for example, and you want to have this constructor, you have to remember to, to call parent construct. And yeah, the, the obvious benefit is that it's less code to write. It's just easy: save, no EntityManager, no custom repositories, you can just go ahead and do things.


And just to wrap things up. So what did we look at? First of all, um, the, the Doctrine EntityManager uses underneath the UnitOfWork which does all the heavy lifting and storing the entity changes until they are persisted. It takes care of the commit order and everything. The proxies for, for finding stuff are useful that it basically allows us to lazily load stuff if we want to, and if not then we can just bypass it by writing custom queries. And the identity map as well for finding stuff is really useful when it comes to minimizing queries and to make sure that we, when we fetch an object from the database and it's something that we worked on before that, that it still has the same state: we work on a consistent object and not like one OrderItem of the name x and one OrderItem of the name y, and then we want to save it, they don't really fit together or the last one wins.

And yeah, that's basically it. So I don't know how much time I have left. I think I rushed through it quite a lot more than I would have to. Um, so yeah, plenty of times for questions I guess.


Yeah. Wait, there's a microphone. Cool, thanks, thanks for catching that, so I don't look bad.

(unaudible) Doctrine to handle concurrent persistent in different processes, the same database?

Um, so, so again, uh, how does Doctrine handle different repositories?

No, if Doctrine is running into 2 processes, php processes?

In different PHP processes. Yeah. So, um, when you have different PHP processes, you will obviously have different unit of works. So there, things will, will not apply, in terms of, you will not have the identity map from one process work with the other process. Um, I'm not sure if there's some common caching you could use, like the l2 caching things like that. I haven't looked into that. Um, but yea, so those things are really, if you work on one php process and, and you just want to access your data there, which is kind of the most common use case. I can't really tell you how it works with multiple PHP process. So, um, everything is good. Um, I think there was someone in the back who had a question or not.

Okay. Uh, since we all have some time, I don't know how eagerly you want to go to, to the jeopardy, I kinda do. But, um, I, I have some code if you kind of feel that this, this was like rushing through everything and just getting a glimpse here and there. If you want, I can give you a quick rundown of the code for, for example, the querying of how it looks in a real project to make it more hands-on and a little bit more approachable. Um, I see someone not? Okay, then let's just do that for, for five minutes. And whoever else wants to leave for, for the, the Jeopardy or just to get some drinks or whatever, feel free to do that.

So only two minutes. Okay then, then we will do it really fast. I, I'm used to that. So I already did this with the talk. So, um, let's look at our list orders controller. So in this case I used an invokable controller. So one controller only has one action, the invoke action. Um, oh, it doesn't, wait a second. Well, since the time it's up anyway. Let's do it differently. Whoever was interested, you can come up and I will, I will show you on my notebook and since I can't get the screen to work. And everyone else, thanks for staying here, for, for listening to my talk. Please give me feedback. Um, and yeah, have a good day and hopefully see you around.