The start of ef always looks good. The problems starts when you have a db with lets say order -> orderrow -> item -> unit relations and start to include. At least in old EF you could easy be in a situation where you load order with rows, but then have a foreach loop getting the items and unit one by one. This because of the lazy loading and when new to EF and working on small datasets it's easy to miss and realise when data grows the bad performance you just built. Also navigation properties that relates to each other and when returned creates circular references. When you know sql it feels like learning ef is like learning sql over again but in new syntax. EF code first also "removes" store procedures. Sometimes you would like these to gain certain performance issues I think. So a follow up guide with more depths and relations between tables and tips to avoid the most common pitfalls would be great!
These are interesting problems. I've never encountered them. I would recommend using explicit includes via directly calling the property, navigation path and all, instead of implicitly including navigation and then blanket pulling data as a 'select *'. You can even use something like automapper to 'project to' which will behave in that same fashion. The circular references, if I'm not mistaken can be avoided by configuring JSON serialization to ignore JSON cycles. Stored procedures can be added to the up/down, but yes it's manually created migration code. Or you can create the stored proc in db and write a call into db context, oooor you could execute raw SQL using 'FromSql' or 'FromSqlRaw'. I created some pretty cool pipelines by rolling my own generic & dynamic procedure call builder, controlling inputs and outputs with a couple DTOs. All in all, I felt EF Core to be complementary to my SQL. SQL/DB development/BI/ETL is all I knew before stumbling into software development and have felt extremely empowered in EF Core since I started using it about 2 years ago.
I create a TableConstraints.cs Add to override On model creating -- TableConstraints.Configure(modelBuilder) if I have a collection of items, say I have a class called 'MenuItem' and this MenuItem has a list of Ingredients which are needed to cook it, my class for MenuItem has: public virtual ICollection Ingredients { get; set; } = new List Then in my ingredients class, simply adding annotation: [Foreign key("MenuItemId")] public virtual MenuItem MenuItem { get; set; } Should do the trick, but that only works when you are using primary keys of each table as the constraint reference. I use alternate keys and have to specify the constraints so Id just add this to my TableConstraints config I mentioned: modelBuilder.Entity() .HasOne(x => x.MenuItem) .WithMany(y => y.Ingredients) .HasForeignKey(x => x.MenuItemId) .HasPrincipalKey(y => y MenuItemId) Can specify here things like: .OnDelete(DeleteBehavior.Restrict) or cascade Or whether property IsRequired. Again, keep in mind that these can pretty much be handled directly in the EF model using annotations but you have more control in the fluent API.
Hi Nick and everybody interested in this topic! First of all, thanks for that video. It's a great starting point for using EF and in general, I really like the topics you cover in your videos. To return to EF, I'd like to ask a question about your oppinion. I found myself preferring EntityTypeConfiguration over adding attributes directly to the entity classes. It's because I have the feeling to have more control over what's created in the database with it. Also the code of the entity classes is more clean - at least in my point of view. What do you think about EntityTypeConfigurations? What do you use in your production code? This question goes out to you all following this video! Thank you for your answers!
EF core or Dapper does not matter but appreciate EF core basics as it helps juniors to start on this journey and give confidence to seniors. Concepts like seeding data, creating indexes with more organised code and unit testing or stress testing might help people going into large scale projects.
Can you show us how to use a sqlproj in a real situation? I believe it was potential to be a great tool, migration can be a nightmare in medium/large projects. I think using dacpac with ef core power tools is more effective, but different opinions makes us better 😊
I had a lot of trouble with PG enums integration with EF core. I ended up running rawsql for any update which had Enum column. What is a best way to integrate it.
Hi Nick, I usually use db.Add(model) since the model is already defined as Movie, rather than db.Movies.Add(model), it is any difference in performance or anything in this practice?
Just in time, with a new project, I'm finally facing the long-feared EF, and so far, so good. But found out some obstacles that, if somebody could help, would be awesome. #1 How do I update only some parts of the entity? Currently, I am fetching the entity, modifying what I need, and then trigger the save. However, that is one more trip (select) needed instead of a simple update. While with db.update(entity), I can avoid it, but that overrides all the entity parts. #2 Having DateTime in the model, how do I update it using DB time (not local time), usually done with a SQL query like NOW() or CURRENT_TIMESTAMP(), but with EF? Any tips are appreciated :)
@@ДмитрийКондратенко-б5ь There can still be a difference. Lets say that the app that calls insert or update is running on the client PC. How do you know he did not tamper with his PC time? Or even if you have control of that app, NTP is disabled for whatever reason, and the time will slide away. With classic SQL NOW() there is no such possibility.
@@martink.7497 , I meant that if you use some parameter in the application with the value of the current time and then want to save it to the database, it is better to do this using the datetime.UtcNow property. In addition, I have never done this, but I have seen similar code, you can for example configure a column in the database using the entity framework so that when a new record is created, the value of a Sql function is written to this column, for example GETUTCDATE(), if we are talking about ms sql server. This will shift the responsibility of determining the current UTC time to the database server. If it is located outside of the user's computer, for example on a server, I think you can quite trust this time value.
@@Cam_all-in-on-GME Not really, he sent the whole entity - id, name and year. EF is just smart about it, that it will update only the changed value, the year in this case, but if, for example, the name was empty, it would overwrite it too. So what if we dont know all the columns and want to update only some? With SQL, just update set column1 = value where id = 1. How to do the same with EF? I tried to pass null values, but it yelled at me that the DB column is not nullable, which is true.
Perfect timing! I’m was just going to learn efcore this afternoon!
Hi Nick! This is a nice one... A follow up would be appreciated! Thanks so much!
I wanted to say the same. I'm using EF Core, but I wasn't aware of "No tracking" which I would have used before for good reasons.
The start of ef always looks good. The problems starts when you have a db with lets say order -> orderrow -> item -> unit relations and start to include. At least in old EF you could easy be in a situation where you load order with rows, but then have a foreach loop getting the items and unit one by one. This because of the lazy loading and when new to EF and working on small datasets it's easy to miss and realise when data grows the bad performance you just built.
Also navigation properties that relates to each other and when returned creates circular references. When you know sql it feels like learning ef is like learning sql over again but in new syntax.
EF code first also "removes" store procedures. Sometimes you would like these to gain certain performance issues I think.
So a follow up guide with more depths and relations between tables and tips to avoid the most common pitfalls would be great!
These are interesting problems. I've never encountered them. I would recommend using explicit includes via directly calling the property, navigation path and all, instead of implicitly including navigation and then blanket pulling data as a 'select *'. You can even use something like automapper to 'project to' which will behave in that same fashion. The circular references, if I'm not mistaken can be avoided by configuring JSON serialization to ignore JSON cycles. Stored procedures can be added to the up/down, but yes it's manually created migration code. Or you can create the stored proc in db and write a call into db context, oooor you could execute raw SQL using 'FromSql' or 'FromSqlRaw'. I created some pretty cool pipelines by rolling my own generic & dynamic procedure call builder, controlling inputs and outputs with a couple DTOs. All in all, I felt EF Core to be complementary to my SQL. SQL/DB development/BI/ETL is all I knew before stumbling into software development and have felt extremely empowered in EF Core since I started using it about 2 years ago.
I'd love to see a deep dive, taking this to the next level. Thanks, Nick. Great content, as usual!
24:18 dbcontext is scoped so the object won't be in memory for future requests
Great intro! I would like suggest how to deal with navigation, foreign keys stuf and etc. It's a hell because the EF's convention 😅
I create a TableConstraints.cs
Add to override On model creating -- TableConstraints.Configure(modelBuilder)
if I have a collection of items, say I have a class called 'MenuItem' and this MenuItem has a list of Ingredients which are needed to cook it, my class for MenuItem has:
public virtual ICollection Ingredients { get; set; } = new List
Then in my ingredients class, simply adding annotation:
[Foreign key("MenuItemId")] public virtual MenuItem MenuItem { get; set; }
Should do the trick, but that only works when you are using primary keys of each table as the constraint reference. I use alternate keys and have to specify the constraints so Id just add this to my TableConstraints config I mentioned:
modelBuilder.Entity()
.HasOne(x => x.MenuItem)
.WithMany(y => y.Ingredients)
.HasForeignKey(x => x.MenuItemId)
.HasPrincipalKey(y => y MenuItemId)
Can specify here things like:
.OnDelete(DeleteBehavior.Restrict) or cascade
Or whether property IsRequired. Again, keep in mind that these can pretty much be handled directly in the EF model using annotations but you have more control in the fluent API.
We need a new performance ef core video.
This was great, thanks for the video!
This is a great intro to efcore. I would like to see the follow-up.
Hi Nick and everybody interested in this topic!
First of all, thanks for that video. It's a great starting point for using EF and in general, I really like the topics you cover in your videos.
To return to EF, I'd like to ask a question about your oppinion.
I found myself preferring EntityTypeConfiguration over adding attributes directly to the entity classes.
It's because I have the feeling to have more control over what's created in the database with it.
Also the code of the entity classes is more clean - at least in my point of view.
What do you think about EntityTypeConfigurations? What do you use in your production code?
This question goes out to you all following this video!
Thank you for your answers!
I'd choose dapper over ef core for the sole reason that I can deploy an AOT compiled binary.
EF core or Dapper does not matter but appreciate EF core basics as it helps juniors to start on this journey and give confidence to seniors. Concepts like seeding data, creating indexes with more organised code and unit testing or stress testing might help people going into large scale projects.
Awesome 😍 keep going I like your content ✨
Wasn't it like a year ago you made a video promoting Dapper over EF? 😅
Anyway, always good to know different ways to solve an issue
Can you show us how to use a sqlproj in a real situation? I believe it was potential to be a great tool, migration can be a nightmare in medium/large projects. I think using dacpac with ef core power tools is more effective, but different opinions makes us better 😊
Could you do pattern designs as a new playlist on your channel?
Also how outdated is Blazor playlist on your channel?
hey nick one question. which frontend framework are you using in dometrain? react?angular? or blazor?
or vue.js maybe?
@@paladincubano dont know xd
I had a lot of trouble with PG enums integration with EF core. I ended up running rawsql for any update which had Enum column.
What is a best way to integrate it.
Why ef core first db call take time how to speed it up , our aws lambda usually timeout
Hi Nick, I usually use db.Add(model) since the model is already defined as Movie, rather than db.Movies.Add(model), it is any difference in performance or anything in this practice?
They’re the same things. One is explicit (db.Movies.Add) and the other is implicit (db.Add). Totally depends on your preference.
Teleio video 👍
I did not see this video on my subscriptions feed. So strange
Just in time, with a new project, I'm finally facing the long-feared EF, and so far, so good.
But found out some obstacles that, if somebody could help, would be awesome.
#1 How do I update only some parts of the entity? Currently, I am fetching the entity, modifying what I need, and then trigger the save. However, that is one more trip (select) needed instead of a simple update. While with db.update(entity), I can avoid it, but that overrides all the entity parts.
#2 Having DateTime in the model, how do I update it using DB time (not local time), usually done with a SQL query like NOW() or CURRENT_TIMESTAMP(), but with EF?
Any tips are appreciated :)
There is like a 'best practice' to store datetime in DB as UTC time always.
@@ДмитрийКондратенко-б5ь There can still be a difference. Lets say that the app that calls insert or update is running on the client PC. How do you know he did not tamper with his PC time? Or even if you have control of that app, NTP is disabled for whatever reason, and the time will slide away. With classic SQL NOW() there is no such possibility.
Your #1 is actually shown in this video where Nick changes the year of the movie and only partially updates the record.
@@martink.7497 , I meant that if you use some parameter in the application with the value of the current time and then want to save it to the database, it is better to do this using the datetime.UtcNow property. In addition, I have never done this, but I have seen similar code, you can for example configure a column in the database using the entity framework so that when a new record is created, the value of a Sql function is written to this column, for example GETUTCDATE(), if we are talking about ms sql server. This will shift the responsibility of determining the current UTC time to the database server. If it is located outside of the user's computer, for example on a server, I think you can quite trust this time value.
@@Cam_all-in-on-GME Not really, he sent the whole entity - id, name and year. EF is just smart about it, that it will update only the changed value, the year in this case, but if, for example, the name was empty, it would overwrite it too. So what if we dont know all the columns and want to update only some? With SQL, just update set column1 = value where id = 1. How to do the same with EF? I tried to pass null values, but it yelled at me that the DB column is not nullable, which is true.
Advanced please.
how does your cli look like that? so pretty
He's probably using Oh My Posh.
Wayyyy too much upfront stuff. Just get into efcore… we likely have environments already and want to see efcore
why is the aspect ratio messed up
What do you mean?
@@nickchapsas it's fine for me (16:9)
@@nickchapsasit was a youtube bug i reopened the video and it was fine