Travis - A few quick responses while I have a couple of minutes on the weekend.
Doing (1) with the MarkLogic Server 3.2 requires having something in front of MLS do URL rewriting into form (2). MarkMail does this using squid, for instance (well, we did, now we do it differently, but you get the drift). Form (2) is supported using the AppServer built-ins (http://developer.marklogic.com/pubs/3.2/apidocs/AppServerBuiltins.html) . So for instance, xdmp:get-request-field() and xdmp:get-request-field-names() are your friends here. Form (3) is more work, and while the server itself does not directly support it, it can probably be done (although I haven't tried it myself). One starting point might be the SOAP Router contributed by the community and found a ways down the page here: http://developer.marklogic.com/code/ Cheers ian > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of > Travis Spencer > Sent: Sunday, May 25, 2008 3:16 PM > To: [email protected] > Subject: [MarkLogic Dev General] Exposing content via Web services > > Good Afternoon All, > > I am trying to figure out the particulars of how one would > create XML Web services to expose the content within MLS. I > would think that this is a very common use case, but I can't > find enough information to actually solve it. If anyone can > point me to documentation, URLs, Webcasts, etc., I would > really appreciate it. Alternatively, can someone help me > fill in the blanks by answering the questions below? > > As I see it, there are three types of interfaces that such > services could have, but I'm stumped on various details with > all of them. I've enumerated these three options below: > > 1. Expose the content using a RESTful Web service API > > To do this, the URLs would have to be unique per resource. > This would mean, for example, that a fictitious user > management system might expose users with URLs like this one: > > http://travisspencer.com/user/tspencer > > To do this, the HTTP app server defined in MLS would have to > map this URL to some XQuery script, e.g., /user.xqy, and pass > it the user name, tspencer, from the URL. I don't see any > way to do this in MLS. Once this mapping was done, assuming > it can be, I'd have to determine the HTTP verb (GET, PUT, > etc.) within the script to know what action to perform. Is > there a custom XQuery function supported by MLS to allow me > to determine the HTTP verb of a request? > > 2. Expose the content using a query API > > To do this, the action and/or parameters would be passed to > my XQuery scripts on the query string like this, for example: > > http://travisspencer.com/user.xqy?action=delete&name=tspencer > > I'm sure that MLS provides a way to pluck out the parameters > and their values from the query string, but I'm not sure how. > Any idea? > > 3. Expose the content using a SOAP-based API > > With SOAP, I'm not even sure where to start. For instance, > how would the WSDL describing the service's interface be > created? Would I have to do it by hand? If so, what about > the non-abstract portion of the contract -- the biding, port, > and service which are specific to the host environment? > Would I have to hand role this too? What if the port or > machine name change? It seems that even these volatile datum > would have to be modified unless MLS is some how able to > automatically generate the WSDL for me. > > If the WSDL is automatically generated, how do I specify > which operations are included, and how do I map those to the > proper XQuery scripts that should be called upon to handle > SOAP messages for those respective operations? Can I > indicate that all the functions in a module, for example, are > to be exposed as one SOAP service and MLS is able to do the > WSDL generation and dispatching of messages? Or would I have > to role the WSDL myself and interrogate the SOAP messages, > determine the operation, and pass it to the appropriate > XQuery script/function? > > What about when a fault occurs? Do I have to create the > entire XML document that represents the SOAP fault myself or > does MLS support some helper XQuery method that I can use in > my catch blocks to create the fault XML? > > Can I only hope for my SOAP Web services to comply with WS-I > BP or is there some way to also support the more advanced > protocols like WS-AT, WS-RM, etc.? Would I have to build > this support myself? > > As you can see, I have many questions about how all this is > done. If I can figure this out though, I'll be ready to do > some really cool stuff w/ MLS. Any help getting to that > point would be much appreciated. > > -- > > Regards, > > Travis Spencer > _______________________________________________ > General mailing list > [email protected] > http://xqzone.com/mailman/listinfo/general > _______________________________________________ General mailing list [email protected] http://xqzone.com/mailman/listinfo/general
