[aforwardslash]

mapping database rows to object structures is an object mapper. An object relational mapper also keeps track of table dependencies (such as related fields).

If you read a row from a database and generate an object whose attributes map the fields in the database and are used to retrieve the values (a data object), that is an object mapper. This means when fetching eg. user.type it will return 1 instead of the data object for the corresponding row on user_type;

If you read a row from a database, exactly like the above, but user.type returns a data object representing the related table row, that's an object relational mapper.

Regarding serialization, why does it matter? Because you need to serialize and de-serialize data objects or models if you're adding cache to eg. a service layer. Also, serialization and de-serialization are quite important when interfacing with eg. external systems - Imagine having an application-wide InvoiceModel that can be transported via REST, GRPC, kafka/json or any other format, and that is database-agnostic.

[zzzeek]

> An object relational mapper also keeps track of table dependencies (such as related fields).

that's not what "relational" means. "relational" means, "a relational database". where we are using SQL statements to deliver data to and from such a database. an ORM that does not directly interpret objects along one-to-many collections /many-to-one attributes is still an object relational mapper.

> Regarding serialization, why does it matter? Because you need to serialize and de-serialize data objects or models if you're adding cache to eg. a service layer.

that's a separate concern from object relational mapping. Do you have the notion that ORMs produce objects that aren't compatible with serialization?

[Arch-TK]

> an ORM that does not directly interpret objects along one-to-many collections /many-to-one attributes is still an object relational mapper.

This is diluting the term ORM beyond any usable definition.

A bit like when people claim that any programming language where you can associate a function with data and have the function implicitly get passed a reference to the data through data.function() call syntax is an OO language.

First off, I question your definition of "object". Is a C struct an "object"? Is a python tuple an "object"? I know that in python it literally is an object, but so is a function. But obviously your definition of an object must be separate from any given language.

To that end, does python's sqlite3 interface constitute an ORM? It converts the results of SQL queries to tuples. What about if I change the row_factory to sqlite3.Row? Is it now an ORM?

Where do you draw the line between something which is and isn't an ORM?

What is the point of such a vague definition?

What isn't an ORM in this case?

[aforwardslash]

> that's not what "relational" means. "relational" means, "a relational database". where we are using SQL statements to deliver data to and from such a database. an ORM that does not directly interpret objects along one-to-many collections /many-to-one attributes is still an object relational mapper.

It doesn't. It means it works as a virtual object database that honors relationships on the underlying schema. The fact that the schema is relational is an implementation detail. Many ORMs will work with non-relational databases, but will honor underlying relations transparently if they exist, such as eg. SqlAlchemy on MongoDB.

An object mapper isn't even a "database-related" technique. Its just a mechanism to map data into objects - ex. JSON deserialization into object attributes is object mapping. When used in conjunction with traditional database mechanisms (such as eg. repository pattern and query builders), can be used to minimize data conversions between systems.

> that's a separate concern from object relational mapping.

Well, it is because in Django ORM or SqlAlchemy you use differerent declarations of data structures for data received from requests, data passed around and data read from and written to the database. It is actually a very leaky implementation where eg. when designing models you feel like you're working on a business object representation, but in fact you're still catering to implementation details of the relational model, such as primary keys, nullable fields, indexes and so on and so on.

> Do you have the notion that ORMs produce objects that aren't compatible with serialization?

That is precisely the point. This is not a feature, it is a quite important limitation. If you use data objects, this issue (as well as complex mapping routines) goes away, because then these objects only hold data and not business logic. These concepts aren't really mainstream in python, but are bread and butter in many other languages.