Sure — here’s an example. Let’s say you have a UploadBlogPost and a ViewBlogPost request. Let’s say the UploadBlogPost request takes an unique id and the post’s body and sticks it into a database. And let’s say ViewBlogPost takes an id, and returns the blog post associated with that id.
The UploadBlogPost request shouldn’t be sent in by just anybody. And what if the unique id isn’t so unique after all? So we write up, let’s say, an AuthenticatedRequest decorator and a TransactionalRequest decorator. The ViewBlogPost request, however, can be sent in by anybody, and it is read-only, so it doesn’t have to be transactional, either. So for UploadBlogPost, you’d have the dependency injection container return a new AuthenticatedRequest(new TransactionalRequest(new UploadBlogPostHandler))); and new ViewBlogPost(); respectively. That’s what I meant by fine-tuned — you do not want none of that if ( $request instanceof UploadBlogPost) { … } else if ( $request instanceof ViewBlogPost ) { … } nastiness. On the other hand, we want to log all requests that comes in, right? And what if that AuthenticatedRequest decorator throws an UserNotAuthenticatedException? We’d want to handle that before that thing bubbles up beyond the router, right? So those are things that apply to the application in its entirety — no matter what request comes in, we’ll always need those middlewares ready for action. That’s where PSR-15 may make more sense. Let’s say we have this in the index.php file — a simplified example, not necessarily correctly implementing PSR-15, but you get the idea: $stack = new Stack(); $stack->add(function ($request, $response) { */ does the logging goodness */}); $stack->add(function ($request, $response) { */ sends the request off to its handler. If the handler doesn’t like it, catch the exception and… */ }); $stack->add(function ($request, $response) { */ … send it to here, where the exception is handled. */ }); It wouldn’t make sense for a framework to provide fine-tuned middleware support, but a ‘one-size-fits-all’ middleware inter-op does make sense. PSR-15 allows a framework to “run” any middleware you may build without a mess of if-else statements. The above example is probably not how I’d write it ‘in the real world’, but I hope you get the idea of what I’m trying to say :) John On Apr 21, 2017, at 1:41 PM, Rasmus Schultz <ras...@mindplay.dk<mailto:ras...@mindplay.dk>> wrote: I'm not sure what you mean by "fine-grained" middleware? This "fine-grained" middleware performs request-to-response processing the same as "general" middleware - they have the same method signature. This is pretty abstract, so can you support this point of view with code-samples or scenario/use-case descriptions to illustrate? On Fri, Apr 21, 2017 at 7:07 PM, Rivera, John <rive...@buffalo.edu<mailto:rive...@buffalo.edu>> wrote: This is exactly how I handle most of my middleware as well — I have a project that uses the CQRS architectural pattern, and I decorate each command handler using this exact pattern. I build the dependency graph in my dependency injection container, and use a mediator to dispatch commands to their (decorated) handlers. This is excellent for fine-tuned control over specific command handlers and their aspects. However, I can see a use for the PSR-15 style middleware — I see two different kinds of middleware: fine-grained middleware (some requests needs to be authenticated, some needs to be within a database transaction, maybe one or two requests result in an event that you need to fire off, etc), and the general middleware (you want to log all requests, you want to catch any exceptions and handle them appropriately, etc). I think the approach we are discussing here is ideal for fine-grained middleware — it makes the dependency graph explicit and easy to configure and maintain separately. PSR-15 is ideal for general middleware, which you can take a more functional approach to — there would (or should) only be a few layers, and they would apply to the application as a whole. Just my two cents. John On Apr 21, 2017, at 12:40 PM, Beau Simensen <simen...@gmail.com<mailto:simen...@gmail.com>> wrote: On Friday, April 21, 2017 at 10:42:11 AM UTC-5, Rasmus Schultz wrote: $kernel = new NotFoundMiddleware(); $kernel = new RouterMiddleware(new Router(...), $kernel); $kernel = new CacheMiddleware($kernel); $kernel = new ErrorHandlerMiddleware(); Off the top of my head, it feels like this would be difficult to automatically wire with a DI container since each middleware will need to be constructed with the previous middleware already instantiated. Especially given you cannot have consistent known arguments, this will be difficult to automate. return $this->container->get($id)->process($request); I think you tried to address this with the ContainerProxyMiddleware but it skips the construction part. How would the container know which $kernel to use in the constructor for the object at $id? This looks nice, API-wise, but it is ignoring how this object would actually be constructed. One of the things we did with Stack was required the first argument to always be the kernel and any additional arguments (if any at all) would have to follow after that. We used Stack Builder to help try and solve those problems. It still seemed messy and I was never very happy with it. -- You received this message because you are subscribed to the Google Groups "PHP Framework Interoperability Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to php-fig+unsubscr...@googlegroups.com<mailto:php-fig+unsubscr...@googlegroups.com>. To post to this group, send email to php-fig@googlegroups.com<mailto:php-fig@googlegroups.com>. To view this discussion on the web visit https://groups.google.com/d/msgid/php-fig/99e602d1-e871-4d19-829a-1330252b508c%40googlegroups.com<https://groups.google.com/d/msgid/php-fig/99e602d1-e871-4d19-829a-1330252b508c%40googlegroups.com?utm_medium=email&utm_source=footer>. For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to a topic in the Google Groups "PHP Framework Interoperability Group" group. To unsubscribe from this topic, visit https://groups.google.com/d/topic/php-fig/B3jtdJA7-6w/unsubscribe. To unsubscribe from this group and all its topics, send an email to php-fig+unsubscr...@googlegroups.com<mailto:php-fig+unsubscr...@googlegroups.com>. To post to this group, send email to php-fig@googlegroups.com<mailto:php-fig@googlegroups.com>. To view this discussion on the web visit https://groups.google.com/d/msgid/php-fig/57FC4BBE-38F8-4B7D-B02D-1E003A0B7D21%40buffalo.edu<https://groups.google.com/d/msgid/php-fig/57FC4BBE-38F8-4B7D-B02D-1E003A0B7D21%40buffalo.edu?utm_medium=email&utm_source=footer>. For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "PHP Framework Interoperability Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to php-fig+unsubscr...@googlegroups.com<mailto:php-fig+unsubscr...@googlegroups.com>. To post to this group, send email to php-fig@googlegroups.com<mailto:php-fig@googlegroups.com>. To view this discussion on the web visit https://groups.google.com/d/msgid/php-fig/CADqTB_imT06jv9v-3DN4_uv1vZmmqO%2B6at8Ny5DpocWNNPmZCg%40mail.gmail.com<https://groups.google.com/d/msgid/php-fig/CADqTB_imT06jv9v-3DN4_uv1vZmmqO%2B6at8Ny5DpocWNNPmZCg%40mail.gmail.com?utm_medium=email&utm_source=footer>. For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "PHP Framework Interoperability Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to php-fig+unsubscr...@googlegroups.com. To post to this group, send email to php-fig@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/php-fig/88F5DE6C-20F1-4CCF-A84C-655FA32AC471%40buffalo.edu. For more options, visit https://groups.google.com/d/optout.