我强烈反对404，而支持204或200的空数据。或者至少应该使用带有404的响应实体。

请求被接收并被正确处理——它确实触发了服务器上的应用程序代码，客户机可能没有犯任何错误，因此整个客户机错误代码(4xx)类可能不合适。

更重要的是，404的发生有很多技术原因。例如，应用程序在服务器上被暂时停用或卸载，代理连接问题等等。

当然，这种情况下存在5xx错误类，但实际上，受影响的中间件组件通常无法知道错误在它们这一边，然后只是假设错误在客户端，然后响应404而不是500/503。

因此，仅根据状态代码，客户端无法区分404(表示“您正在寻找的东西不存在”)和404(表示“有严重错误，请将此错误报告给运维团队”)。

This can be fatal: Imagine an accounting service in your company that lists all the employees that are due to an annual bonus. Unfortunately, the one time when it is called it returns a 404. Does that mean that no-one is due for a bonus, or that the application is currently down for a new deployment and the 404 is actually coming from the tomcat that it's supposed to be installed into, instead of from the application itself? These two scenarios yield the same status code, but they are fundamentally different in their meaning.

对于需要知道所请求的资源不存在而不是暂时不可访问的应用程序来说，没有响应实体的404几乎是行不通的。

此外，许多客户端框架通过抛出异常来响应404，而不询问进一步的问题。这迫使客户端开发人员捕获异常，对其进行评估，然后基于此决定是否将其记录为由监视组件捕获的错误，或者是否忽略它。这对我来说也不太好。

The advantage of 404 over 204 is that it can return a response entity that may contain some information about why the requested resource was not found. But if that really is relevant, then one may also consider using a 200 OK response and design the system in a way that allows for error responses in the payload data. Alternatively, one could use the payload of the 404 response to return structured information to the caller. If he receives e.g. a html page instead of XML or JSON that he can parse, then that is a good indicator that something technical went wrong instead of a "no result" reply that may be valid from the caller's point of view. Or one could use a HTTP response header for that.

尽管如此，我还是更喜欢204或200的空白回复。这样，请求的技术执行状态就与请求的逻辑结果分开了。2xx的意思是“技术执行ok，这就是结果，处理它”。

我认为在大多数情况下，应该让客户来决定一个空的结果是否可以接受。通过返回404而不返回响应实体(尽管技术执行正确)，客户端可能决定将根本不是错误的情况视为错误。

Another perspective: From an operations point of view a 404 may be problematic. Since it can indicate a connectivity/middleware problem rather than a valid service response, i would not want a fluctuating number of "valid" 404s in my metrics/dashboards that might conceal genuine technical issues (e.g. a misconfigured proxy somewhere in the request routing) that should be investigated and fixed. This is further excarbated by some APIs even using 404 instead of 401/403 (e.g. gitlab does such a thing), to conceal the information that the request URI would have been valid but the request lacked authorization to access it. In this case too a 404 should be treated as a technical error and not as a valid "resource not found" result.

Edit: Wow, this has caused a lot of controversy. Here is another argument against 404: Strictly from a HTTP spec (RFC7231) point of view, 404 does not even mean that a resource does not exist. It only means that the server has no current representation of the requested resource available, and this even may be only temporary. So strictly by HTTP spec, 404 is inherently unreliable regarding the nonexistence of a requested thing. If you want to communicate that the requested thing positively does not exist, do not use 404.

起初，我认为204是有意义的，但经过讨论，我相信404是唯一真正正确的回答。考虑以下数据:

用户:约翰，彼得

METHOD  URL                      STATUS  RESPONSE
GET     /users                   200     [John, Peter]
GET     /users/john              200     John
GET     /unknown-url-egaer       404     Not Found
GET     /users/kyle              404     User Not found
GET     /users?name=kyle`        200     []
DELETE  /users/john              204     No Content

背景知识:

the search returns an array, it just didn't have any matches but it has content: an empty array. 404 is of course best known for url's that aren't supported by the requested server, but a missing resource is in fact the same. Even though /users/:name is matched with users/kyle, the user Kyle is not available resource so a 404 still applies. It isn't a search query, it is a direct reference by a dynamic url, so 404 it is. After suggestions in the comments, customizing the message of the 404 is another way of helping out the API consumer to even better distinguish between complete unknown routes and missing entities.

不管怎样，我的意见。

在以前的项目中，我使用了404。如果没有用户9，则没有找到对象。因此404 Not Found是合适的。

如果对象存在，但没有数据，则为204 no Content。我认为在你的例子中，这个物体并不存在。

如果仅仅因为没有响应数据而返回404，则任何客户端都会感到非常困惑。

对我来说，响应代码200和一个空的主体足以理解一切都是完美的，但没有数据匹配的要求。

看完疑问后，为什么不应该使用404 ?

根据RFC 7231，正确的状态码是204

在上面的回答中，我注意到一个小错误:

1.—资源为:/users

2.- /users/8不是资源，而是:路由参数为8的资源/users，消费者可能注意不到，也不知道区别，但是发布者知道，而且必须知道!所以他必须为消费者返回一个准确的响应。时期。

so:

基于RFC: 404是不正确的，因为找到了资源/用户，但是使用参数8执行的逻辑没有找到任何内容作为响应返回，因此正确的答案是:204

这里的要点是:404甚至没有找到处理内部逻辑的资源

204是a:我找到了资源，逻辑被执行了，但我没有发现任何数据使用你在路由参数中给出的标准，所以我不能返回任何东西给你。对不起，核实你的标准后再打电话给我。

200:好吧，我找到了资源，逻辑被执行(即使当我不被迫返回任何东西)采取这一点，并在你的意愿使用它。

205:(GET响应的最佳选项)我找到了资源，逻辑被执行了，我有一些内容给你，好好使用它，哦，顺便说一下，如果你要在视图中共享这个，请刷新视图以显示它。

希望能有所帮助。

现有的答案没有详细说明使用路径参数还是查询参数是有区别的。

In case of path parameters, the parameter is part of the resource path. In case of /users/9, the response should be 404 because that resource was not found. /users/9 is the resource, and the result is unary, or an error, it doesn't exist. This is not a monad. In case of query parameters, the parameter is not part of the resource path. In case of /users?id=9, the response should be 204 because the resource /users was found but it could not return any data. The resource /users exists and the result is n-ary, it exists even if it is empty. If id is unique, this is a monad.

使用路径参数还是查询参数取决于用例。我更喜欢将路径参数用于强制的、规范的或标识参数，将查询参数用于可选的、非规范的或属性参数(如分页、排序区域设置等)。在REST API中，我会使用/users/9而不是/users?Id =9，特别是因为可能嵌套获取“子记录”，如/users/9/ SSH -keys/0获取第一个公共SSH密钥或/users/9/address/2获取第三个邮政地址。

我更喜欢使用404。原因如下:

Calls for unary (1 result) and n-ary (n results) methods should not vary for no good reason. I like to have the same response codes if possible. The number of expected results is of course a difference, say, you expect the body to be an object (unary) or an array of objects (n-ary). For n-ary, I would return an array, and in case there are not results, I would not return no set (no document), I would return an empty set (empty document, like empty array in JSON or empty element in XML). That is, it's still 200 but with zero records. There's no reason to put this information on the wire other than in the body. 204 is like a void method. I would not use it for GET, only for POST, PUT, and DELETE. I make an exception in case of GET where the identifiers are query parameters not path parameters. Not finding the record is like NoSuchElementException, ArrayIndexOutOfBoundsException or something like that, caused by the client using an id that doesn't exist, so, it's a client error. From a code perspective, getting 204 means an additional branch in the code that could be avoided. It complicates client code, and in some cases it also complicates server code (depending on whether you use entity/model monads or plain entities/models; and I strongly recommend staying away from entity/model monads, it can lead to nasty bugs where because of the monad you think an operation is successful and return 200 or 204 when you should actually have returned something else). Client code is easier to write and understand if 2xx means the server did what the client requested, and 4xx means the server didn't do what the client requested and it's the client's fault. Not giving the client the record that the client requested by id is the client's fault, because the client requested an id that doesn't exist.

最后一点:一致性

GET /用户/ 9 PUT /users/9和DELETE /users/9

PUT /users/9和DELETE /users/9已经必须在成功更新或删除的情况下返回204。如果用户9不存在，它们应该返回什么?根据所使用的HTTP方法将相同的情况显示为不同的状态代码是没有意义的。

Besides, not a normative, but a cultural reason: If 204 is used for GET /users/9 next thing that will happen in the project is that somebody thinks returning 204 is good for n-ary methods. And that complicates client code, because instead of just checking for 2xx and then decoding the body, the client now has to specifically check for 204 and in that case skip decoding the body. Bud what does the client do instead? Create an empty array? Why not have that on the wire, then? If the client creates the empty array, 204 is a form of stupid compression. If the client uses null instead, a whole different can of worms is opened.