Thanks a lot for your help, your answers clarified many things for me Best Regards,
Diana. On 30 nov. 2011, at 15:26, Sergey Beryozkin wrote: > On 30/11/11 14:05, Diana ALLAM wrote: >> Hi, >> >> On 30 nov. 2011, at 11:59, Sergey Beryozkin wrote: >> >>> Hi >>> On 30/11/11 09:29, Diana ALLAM wrote: >>>> Hello, >>>> thank you for your reply. >>>> >>>> Currently, I am doing a study about well dynamic typing of web service and >>>> I choose cxf >>>> to run some examples. >>>> In fact, for SOAP/WS* services, there is a standard used , wsdl files, >>>> which are generated by a web service. >>>> A client can use this wsdl to generate an appropriated interface in java >>>> to build an appropriated request to >>>> the designed web service. Thus, I can say, following this way, there must >>>> not be any typing errors. >>>> But for Restful services, the situation differs a bit. There is a WADL >>>> files to describe the web services like the >>>> WSDL files in SOAP/WS*, but WADL still not a standard. >>> >>> it's a W3C submission and it appears users who work with RESTful services >>> and happen to see the benefit of providing the documents such as WADL just >>> use WADL; >> >> Yes, but I mean that WADL is not yet used by all Restful services like the >> WSDL for WS* >>> >>>> My question is, if a restful web service doesn't define a WADL, thus the >>>> client can't generate in java the appropriated >>>> classes (like using wadltojava), then how he can know the type of messages? >>>> I suppose that in the absence of a WADL for a web service, there is a text >>>> description file describing all methods and >>>> there messages types. Thus, the client has to read this document to know >>>> how to build his messages. >>>> Here, he can make mistakes by sending to a restful resource an >>>> inappropriate data. >>>> >>>> Is my reasoning correct about that? >>> That may not be that big an issue actually - unless we are talking about >>> the completely generic clients, but indeed, with the fluent HTTP-centric >>> API one just needs to apply a bit more care to making sure the correct >>> messages are sent and such - but as I said if the code >>> were not corect it work it would fail at the first try, then the user would >>> go and check the SLA or whatever and update the code. >>> In CXF we support both styles: >>> http://cxf.apache.org/docs/jax-rs-client-api.html >>> so choose the option you like or experiment with both ones >>> >>> Proxies can also be generated from the existing WADLs: >>> http://cxf.apache.org/docs/jaxrs-services-description.html >>> >>> even WSDLs can be used as the input (this option is a bit more involved) >> >> Thank you a lot for these informations. >> I can deduce that from the client developer side, calling a WS* web service >> is more standardized than a restful one, >> because the user can get a WSDL and creates properly a proxy, while for Rest >> it will not be the case if >> no Wadl is provided. >> >> In your opinion, isn't it an advantage for WS*? Doesn't it an annoying >> checking problem for restful's clients? > > What we do in CXF is we keep our heads down and make sure people can write > WS-* and RESTful services the way they need to, that is our philosophy, > because in production the issues like styles may not necessarily take a > precedence, there are many different factors which can tilt a decision into > going with SOAP only, REST only, SOAP & REST; you will see me answering SOAP > questions if I happen to remember something :-). I'd suggest to take this > particular discussion offline if you don't mind. > >> Will WADL be defined for all Restful services in the coming days, such as >> WSDL for WS* services today? >> > > Clearly, it's very important for many users that something like WADL has a > final standard tag, I'm not sure right now if it will make it into a final > spec but even if it will some users would still prefer a fluent HTTP-centric > API and won't rely on WADL at all > > Cheers, Sergey > >> Regards, >> >> Diana >> >> >>> Sergey >>>> >>>> >>>> Regards, >>>> >>>> Diana >>>> >>>> On 29 nov. 2011, at 18:01, Sergey Beryozkin wrote: >>>> >>>>> Hi >>>>> >>>>> On 29/11/11 15:33, Diana ALLAM wrote: >>>>>> Hello, >>>>>> >>>>>> After using the restful examples in the "samples" directory of cxf >>>>>> distribution, I distinguished >>>>>> two types of restful request for "put/post" CRUD methods. >>>>>> 1) the first one is by sending a request with an xml object description >>>>>> (samples -> jax_rs -> basic) >>>>>> Follows, there is a part of the code example for calling the put method >>>>>> by using an input file containing the "customer" xml code: >>>>>> // Sent HTTP PUT request to update customer info >>>>>> System.out.println("\n"); >>>>>> System.out.println("Sent HTTP PUT request to update customer >>>>>> info"); >>>>>> Client client = new Client(); >>>>>> String inputFile = >>>>>> client.getClass().getResource("update_customer.xml").getFile(); >>>>>> URIResolver resolver = new URIResolver(inputFile); >>>>>> File input = new File(resolver.getURI()); >>>>>> PutMethod put = new >>>>>> PutMethod("http://localhost:8080/CustomerRestfullTest/jaxrs/customerservice/customers";); >>>>>> RequestEntity entity = new FileRequestEntity(input, "text/xml; >>>>>> charset=ISO-8859-1"); >>>>>> put.setRequestEntity(entity); >>>>>> HttpClient httpclient = new HttpClient(); >>>>>> >>>>>> try { >>>>>> int result = httpclient.executeMethod(put); >>>>>> System.out.println("Response status code: " + result); >>>>>> System.out.println("Response body: "); >>>>>> System.out.println(put.getResponseBodyAsString()); >>>>>> >>>>>> System.out.println(put.getResponseHeader(Message.ENDPOINT_ADDRESS)); >>>>>> } finally { >>>>>> // Release current connection to the connection pool once >>>>>> you are >>>>>> // done >>>>>> put.releaseConnection(); >>>>>> } >>>>>> >>>>> >>>>> This is a very old code that will be replaced >>>>> >>>>>> 2) the second one is by using an instance of a Java class, here is >>>>>> Customer, instead of getting data from an input file (samples -> >>>>>> jax_rs -> basic_https) >>>>>> >>>>>> WebClient wc = WebClient.create(BASE_SERVICE_URL, >>>>>> CLIENT_CONFIG_FILE); >>>>>> Customer customer = new Customer(); >>>>>> customer.setId(123); >>>>>> customer.setName("Mary"); >>>>>> Response resp = wc.put(customer); >>>>>> >>>>>> >>>>>> I prefer to use the second method because in the first method, >>>>>> a developer could make mistakes in his input file, then the server will >>>>>> reply by a fault message, while the second >>>>>> method allows to avoid such errors as there is a customer object conform >>>>>> to a specified class Customer. >>>>>> >>>>>> I think also the second method is more adapted if the client wishes >>>>>> using a wadl file, then by using wadltojava, all the appropriated classes >>>>>> are generated (like the Customer class in my previous example). >>>>> Agreed >>>>>> But my question is, when and why a developer need to use the first >>>>>> method? >>>>> >>>>> If developers do prefer using Apache HttpClient for whatever reasons then >>>>> they may want to use that API >>>>> >>>>> Cheers, Sergey >>>>> >>>>>> And by such a use, the developer doesn't take the risk of sending >>>>>> inadapted data to the server? >>>>>> and how the developer knows which data description form (for the >>>>>> customer example, it is an xsd) he must use in this case, if he doesn't >>>>>> have a wadl? >>>>>> because I suppose that if he had a wadl then he must use the second >>>>>> method. >>>>>> >>>>>> Thank you in advance. >>>>>> >>>>>> Regards, >>>>>> >>>>>> Diana >>>>>> >>>>>> >>>>> > > > -- > Sergey Beryozkin > > Talend Community Coders > http://coders.talend.com/ > > Blog: http://sberyozkin.blogspot.com
