当没有找到 REST 资源时返回404是否正确？

A 404 response in this case is pretty typical and easy for API users to consume.

One problem is that it is difficult for a client to tell if they got a 404 due to the particular entity not being found, or due to a structural problem in the URI. In your example, /foos/5 might return 404 because the foo with id=5 does not exist. However, /food/1 would return 404 even if foo with id=1 exists (because foos is misspelled). In other words, 404 means either a badly constructed URI or a reference to a non-existent resource.

Another problem arises when you have a URI that references multiple resources. With a simple 404 response, the client has no idea which of the referenced resources was not found.

Both of these problems can be partially mitigated by returning additional information in the response body to let the caller know exactly what was not found.

Yes, it is pretty common to return 404 for a resource not being found. Just like a web page, when it's not found, you get a 404. It's not just REST, but an HTTP standard.

Every resource should have a URL location. URLs don't need to be static, they can be templated. So it's possible for the actual requested URL to not have a resource. It is the server's duty to break down the URL from the template to look for the resource. If they resource doesn't exist, then it's "Not Found"

Here's from the HTTP 1.1 spec

404 Not Found

The server has not found anything matching the Request-URI. No indication is given of whether the condition is temporary or permanent. The 410 (Gone) status code SHOULD be used if the server knows, through some internally configurable mechanism, that an old resource is permanently unavailable and has no forwarding address. This status code is commonly used when the server does not wish to reveal exactly why the request has been refused, or when no other response is applicable.

Here's for 204

204 No Content

The server has fulfilled the request but does not need to return an entity-body, and might want to return updated metainformation. The response MAY include new or updated metainformation in the form of entity-headers, which if present SHOULD be associated with the requested variant.

If the client is a user agent, it SHOULD NOT change its document view from that which caused the request to be sent. This response is primarily intended to allow input for actions to take place without causing a change to the user agent's active document view, although any new or updated metainformation SHOULD be applied to the document currently in the user agent's active view.

The 204 response MUST NOT include a message-body, and thus is always terminated by the first empty line after the header fields.

Normally 204 would be used when a representation has been updated or created and there's no need to send an response body back. In the case of a POST, you could send back just the Location of the newly created resource. Something like

@POST
@Path("/something")
@Consumes(...)
public Response createBuzz(Domain domain, @Context UriInfo uriInfo) {
int domainId = // create domain and get created id
UriBuilder builder = uriInfo.getAbsolutePathBuilder();
builder.path(Integer.toString(domainId));  // concatenate the id.
return Response.created(builder.build()).build();
}

The created(URI) will send back the response with the newly created URI in the Location header.

Adding to the first part. You just need to keep in mind that every request from a client is a request to access a resource, whether it's just to GET it, or update with PUT. And a resource can be anything on the server. If the resource doesn't exist, then a general response would be to tell the client we can't find that resource.

To expand on your example. Let's say FooService accsses the DB. Each row in the database can be considered a resource. And each of those rows (resources) has a unique URL, like foo/db/1 might locate a row with a primary key 1. If the id can't be found, then that resource is "Not Found"

A 4XX error code means error from the client side.
As you request a static resource as an image or a html page, returning a 404 response makes sense as :

The HTTP 404 Not Found client error response code indicates that the server can't find the requested resource. Links which lead to a 404 page are often called broken or dead links, and can be subject to link rot.

As you provide to clients some REST methods, you rely on the HTTP methods but you should not consider REST services as simple resources.
For clients, an error response in the REST method is often handled close to errors of other processings.

For example, to catch errors during REST invocations or somewhere else, clients could use catchError() of RxJS.

We could write a code (in TypeScript/Angular 2 for the sample code) in this way to delegate the error processing to a function :

return this.http
.get<Foo>("/api/foos")
.pipe(
catchError(this.handleError)
)
.map(foo => {...})

The problem is that any HTTP error (5XX or 4XXX) will terminate in the catchError() callback.
It may really make the REST API responses misleading for clients.

If we do a parallel with programming language, we could consider 5XX/4XX as exception flow.
Generally, we don't throw an exception only because a data is not found, we throw it as a data is not found and that that data would have been found.
For the REST API, we should follow the same logic.

If the entity may not be found, returning OK in the two cases is perfectly fine :

@GET
@Path("/{fooId}")
@Produces(MediaType.APPLICATION_XML)
public Response getFoo(@PathParam("fooId") final String fooId)
throws IOException, ParseException {
final Foo foo = fooService.getFoo(fooId);


if (foo != null){
return Response.status(Response.Status.OK).entity(foo).build();
}


return Response.status(Response.Status.OK).build();


}

The client could so handle the result according to the result is present or missing.
I don't think that returning 204 brings any useful value.
The HTTP 204 documentation states that :

The client doesn't need to go away from its current page.

But requesting a REST resource and more particularly by a GET method doesn't mean that the client is about terminating a workflow (that makes more sense with POST/PUT methods).

The document adds also :

The common use case is to return 204 as a result of a PUT request, updating a resource, without changing the current content of the page displayed to the user.

We are really not in this case.

Some specific HTTP codes for classical browsing matche finely with return codes of REST API (201, 202, 401, and so for...) but this is not always the case. So for these cases, rather than twisting original codes, I would favor to keep them simple by using more general codes : 200, 400.

Though this question already have an accepted answer, I believe it's really an opinionated thing. Adding my two cents to help you make a more informed decision about the response code.

404 - Not Found. (Reference)

The origin server did not find a current representation for the target resource or is not willing to disclose that one exists.

The resource may exist and you may not have permission to see the resource, will also be equivalent of Not Found. So 404 for a call where data doesn't exist is a very apt thing to do.

Now as for a non-existing URL; though 404 is a widely adapted response code 400 is a more appropriate code.

400 - Bad Request (Reference)

The server cannot or will not process the request due to something that is perceived to be a client error (e.g., malformed request syntax, invalid request message framing, or deceptive request routing).

If you put an invalid parameter in the request, what would be the response code? If query param has a typo, what should be response code?

Answer to both is 400.

Most of the file-servers, return 404 for invalid URL because for an invalid URL they try to look for a file, which they can't find on the storage ~= Resource Not Found

Apart from the HTTP Status Code, the response will have some info about the error details, where one can be more descriptive about the error and can clear the ambiguity.

If client is calling with an invalid URL, it's an integration issue and should be caught at least during the sanity. No-way they will push the code to production without testing and catching this. Even if they do, God bless them!

tl;dr - 404 for not-found resource; 400 for not-found URL.

Edit (Dec-22)