As i grow in my professional career i consider naming conventions very important. I noticed that people throw around controller, LibraryController, service, LibraryService, and provider, LibraryProvider and use them somewhat interchangeable. Is there any specific reasoning to use one vs the other?
If there are websites that have more concrete definitions that would be great.
In Java Spring, Spring Boot and also .NET you would have:
Repository: persist data in the database and perform SQL queries.Service: contain most of the business logicController: define REST endpoints, which contains as little logic as possible.Conceptually this means that the WHAT (functional) is separated from the HOW (technical) as much as possible. The services try to stay technologically neutral. By contrast a controller only wants to define an external contract for communication. And finally the repository only wants to facilitate the access to the database.
Organizing your code in this way keeps the business logic short, clean and maintainable. Unfortunately it is not always easy to keep them separated. e.g. It is tempting to pollute or enrich your objects with meta-data in the form of decorators/annotations. (e.g. database column name and data type).
Some developers don't see harm in this and get away with it. Others keep their objects strictly separated and define multiple sets of objects.
Having multiple objects means that you need Mappers to convert one type of object to another. Some frameworks can do this for you (e.g. MapStruct).
It would be easy to claim that strictness is always a good thing, but it can slow you down. It's okay to strike a balance.
In Node.js, the concepts of controllers and services are identical. However the term Repository isn't used very often. Instead, they would call that a Provider or sometimes they would just generalize Repositories as a kind of Service.
NestJS has stronger opinions about this (which can be a good thing). The naming conventions of NestJS (a Node.js framework) are strongly influenced by the naming conventions of Angular, which is of course a totally different kind of framework (front-end).
(For completeness, in Angular, a Provider is actually just something that can be injected as a dependency. Most providers are Services, but not necessarily. (It could be a Guard or even a Module. A Module would be more like a set of tools/driver or a connector.)
PS: Anyway, the usage of the term Module is a bit confusing because there also are "ES6 modules", which is a totally different thing.)
ES6 and more modern version of javascript (including typescript) are extremely powerful when it comes to (de)constructing objects. And that makes mappers unnecessary.
Having said that, most Node.js and Angular developers prefer to use typescript these days, which has more features than java or C# when it comes to defining types.
So, all these frameworks are influencing each other. And they pretty much all agree on what a Controller and a Service is. It's mostly the Repository and Provider words that have different meanings. It really is just a matter of conventions. If your framework has a convention, then stick to that. If there isn't one, then pick one yourself.
These terms can be synonymous with each other depending on context, which is why each framework or language creator is free to explicitly declare them as they see fit... think function/method/procedure or process/service, all pretty much the same thing but slight differences in different contexts.
Just going off formal English definitions:
These definitions are just listed to the explain how the developer looks at common English meanings when defining the terminology of a framework or language; it's not always one for one and the similarity in terminology actually provides the developer with a means of naming things that are very very similar but still are slightly different.
So for example, lets take AngularJS. Here the developers decided to use the term Controller to imply "HTML Controller", a Service to imply something like a "Quasi Class" since they are instantiated with the New keyword and a Provider is really a super-set of Service and Factory which is also similar. You could really program any application using any of them and really wouldn't lose anything much; though one might be a little better than another in certain context, I don't personally believe its worth the extra confusion... essentially they are all providers. The Angular people could have just defined factory, provider and service as a single term "provider" and then passed in modifiers for things like "static" and "void" like most languages and the exact same functionality could have been provided; this would have been my preference, however I've learned not to fight the conventions and terminology of the frameworks your working no matter how strongly you disagree.
Old dev here that stumbled on this. My opinion and how I’ve seen it used over the last 20 years shows that it varies by language but the Java C# crowd mostly uses them as follows.
A service handles business logic and deals with domain objects. You find services in controllers and other services.
A repository does NOT handle business logic, but instead acts like a pool of domain objects (with helper methods for finding or persisting them. Services often contain repositories. Repositories often contain a context and are responsible for mapping from infrastructure shaped data to domain shaped data if the definitions have drifted apart. Controllers also often contain repositories for crud endpoints.
A context handles infrastructure the domain owns. Most often this is a database, but context means that anything that touches this data does so through (in) this context. A context returns infrastructure shaped data. A repository often contains a context. Context directly in services is sometimes appropriate. Context in controller is a hard no.
A provider provides access to infrastructure some other app owns. Most often these are rest apis, but can also be kafka streams or rpc classes that read data from or push data to someone else. If the source of truth for some of your domain objects fields changes you will probably see a provider next to a context in your repository, and your repository handles insulating the rest of your code from that change. Providers that provide rpc functionality are often found in services. In micro services or gateways or vertical slice architecture you sometimes see providers directly in controllers.
One old guy’s opinion but I hope it helps.
