ok, we were on the same page from the beginning so. sorry for unprecise wording.
On Tue, Jul 22, 2008 at 2:25 PM, Assaf Arkin <[EMAIL PROTECTED]> wrote: > On Tue, Jul 22, 2008 at 12:09 PM, Alexis Midon <[EMAIL PROTECTED]> wrote: > > The initial question was: should the REST connector use element (of a > simple > > type) or simple type for header parts in the wsdl? > > > > <part name="headerA" type="xs:string"/> > > or > > <xs:element type="xs:string" name="foo" /> > > ... > > <part name="headerA" element="foo"/> > > > > So in the message, I could get: > > <message> > > <headerA>a-string</headerA> > > or > > <headerA><foo>a-string</foo></headerA> > > </message> > > > > My goal was to support both. If the part element has a child element > > (<foo>), get its text content, else get the text content of the part > > element. If none thow an exception. > > (not a really deep recursion) > > > > Using elements make the life easier on the designer side -- every part > will > > have an element (but it does not mean that the user must provide one) -- > and > > allows to use default values. But for hand-written wsdls, having an > element > > for every simple type is quite cumbersome. > > > > So the idea was to support both. Which will you go for? > > Both! > > I was specifically commenting on: > > "take the content of the first encountered text node." > > That would fail in two case. First, the simple type element could > have multiple text nodes (something to do with entity expansion), so > you're only reading half the value. Second, it's actually a complex > type and instead of detecting and failing, you're silently swallowing > the error. > > <foo> > <bar>bad!</bar> > </foo> > > You end up reading the first encountered text node, which is the > whitespace leading up to the bar element, instead of detecting and > failing because foo is not a simple type. > > Assaf > > > > > > > > > > On Tue, Jul 22, 2008 at 11:21 AM, Assaf Arkin (JIRA) <[EMAIL PROTECTED]> > > wrote: > > > >> > >> [ > >> > https://issues.apache.org/jira/browse/ODE-344?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12615710#action_12615710 > ] > >> > >> Assaf Arkin commented on ODE-344: > >> --------------------------------- > >> > >> Mixed content is the root of all unexplained bugs. > >> > >> If the element is simple type, we can say that mapping that element is > the > >> same as mapping its content (i.e. all its text nodes). If the element > is > >> complex type, something is wrong, throw a selection fault. If someone > >> really meant to discard the element contents except for the text nodes, > they > >> can say so explicitly using XPath. > >> > >> > Support element for headers and query strings > >> > --------------------------------------------- > >> > > >> > Key: ODE-344 > >> > URL: https://issues.apache.org/jira/browse/ODE-344 > >> > Project: ODE > >> > Issue Type: Improvement > >> > Components: Axis2 Integration > >> > Reporter: Alexis Midon > >> > > >> > So far, parts bound to HTTP headers or encoding withe > >> url-encoded/x-www-form-urlencoded must be simple types. > >> > This improvement aims at support elements too. > >> > So for a given part A, if A as a text-node, take the content of that > node > >> ; > >> > else check if a child element exists and take the text content of that > >> element. > >> > Which can be summed up in: take the content of the first encountered > text > >> node. > >> > >> -- > >> This message is automatically generated by JIRA. > >> - > >> You can reply to this email to add a comment to the issue online. > >> > >> > > >
