Is option 2 such as bad thing?
But here's a proposed work around -- define a Java object that is an
array of NamedValue objects (e.g., suboptions) and use that object as
your input parameter rather than specifying an array in the request:
NamedValue[] suboptions = new NamedValue[]
{
new NamedValue("dummy2-1", "val2-1"),
new NamedValue("dummy2-2", new Integer(314))
};
NamedValue[] options = new NamedValue[]
{
new NamedValue("dummy1", "dummy_val1"),
new NamedValue("dummy2", suboptions)
};
Anne
On 4/13/05, Tim K. (Gmane) <[EMAIL PROTECTED]> wrote:
> Anne,
>
> That makes sense. That's why I kept pushing and asking questions just to
> make sure what looks like common sense is actually true.
>
> So it looks like I would have at least 3 choices here:
>
> 1) Define in the schema all possible types ArrayOfXxxx that could occur at
> run-time. This is less than ideal because errors would not be discovered
> until run-time, but it could be done with enough testing. Also, the Axis bug
> would still need to be fixed so that it works at all.
>
>
> 2) Change the definition of the value element of NamedValue to be:
>
> <xsd:element name="value" type="xsd:anyType" maxOccurs="unbounded"/>
>
> But then end up with a generated NamedValue class on the client side whose
> value is an Object[] instead of Object. The server class would also have to
> be changed accordingly and that's not really an option at this point as
> there is *a lot* of code written and it's too late to change it. This is
> more of a hack than a real solution.
>
>
> 3) Use rpc/encoded instead which I know works well, except for custom
> exception/fault deserialization in .NET Maybe something custom can be done
> on the .NET side for this case.
>
>
>
> Tim
>
>
> Anne Thomas Manes wrote:
> Well -- actually, you shouldn't be able to send an array of String
> unless you had previously defined a type which is an array of string.
> You see -- that what's really going wrong in Tim's application.
>
> Consider this point: The "value" element has not been defined to allow
> multiple occurrences. To make that work, then you would need to define
> the "value" element like this:
>
> <xsd:element name="value" type="xsd:anyType" maxOccurs="unbounded"/>
>
> But the way it's defined, an <options> element contains one <name>
> element and one <value> element.
>
> You see, this is the really challenging thing about using
> <xsd:anyType>. Axis can't generate schema type definitions on the fly
> (or if it did, it would have to send the schema with the request).
>
> Assuming that you had previously defined an ArrayOfString type:
>
> <xsd:complexType name="ArrayOfString"
> <xsd:sequence>
> <xsd:element name="item" type="xsd:string" maxOccurs="unbounded"/>
> <xsd:sequence>
> <xsd:complexType>
>
> Then you should be able to send:
>
> NamedValue object:
> name: "dummy2"
> value: String[]
>
> And on the wire, it would look like this:
>
> <options>
> <name>dummy2</name>
> <value xsi:type="ns1:ArrayOfString">
> <item>blah</item>
> <item>blah</item>
> </value>
> </options>
>
>
>
> On 4/13/05, Tim K. (Gmane) <[EMAIL PROTECTED]> wrote:
>
>
> Hi Anne,
>
> I would be curious to see what the correct message should look like for a
> String[] as the value of the NamedvAlue, i.e. something like this:
>
> NamedValue object:
> name: "dummy-2"
> value: String[]
>
>
> <options>
> <name>dummy2</name>
> <value xsi:type=".....">
> <!-- How do you form the String[] in here? -->
> </value>
> </options>
>
> Are wrappers always needed in this case?
>
> Thanks.
>
> Tim
>
>
> Anne Thomas Manes wrote:
> You're right. The array within the array should be mapped to this:
>
> <login xmlns="http://some/namespace";>
> <username>tim</username>
> <password>tim</password>
> <options>
> <name>dummy1</name>
> <value xsi:type="xsd:string">dummy_val1</value>
> </options>
> <options>
> <name>dummy2</name>
> <value xsi:type="ns1:NamedValue" xmlns:ns1="http://some/namespace";>
> <name>dummy2-1</name>
> <value xsi:type="xsd:string">val2-1</value>
> <name>dummy2-2</name>
> <value xsi:type="xsd:int">314</value>
> </value>
> </options>
> </login>
>
> I suggest you file a bug report, because Axis is not generating the
> right message structure per the WSDL.
>
> Anne
>
> On 4/13/05, Tim K. (Gmane) <[EMAIL PROTECTED]> wrote:
>
>
> Hi Anne,
>
> I actually started from Java classes using Java2WSDL (I know this is not
> the best practice, but if there's something wrong with the WSDL I could try
> and fix it by hand).
>
> The ArrayOfNamedValue is defined like below, but it does not seem to come
> into play at all:
>
>
>
> - <complexType name="ArrayOfNamedValue">
>
>
> - <sequence>
>
>
> <element name="item" type="impl:NamedValue" minOccurs="0"
> maxOccurs="unbounded" />
> </sequence>
> </complexType>
> I exposed this method:
> NamedValue[] login(String username, String password, NamedValue[] options)
> Which in the WSDL the request looks like this:
>
>
>
> - <element name="login">
>
>
> - <complexType>
>
>
> - <sequence>
>
>
> <element name="username" type="xsd:string" />
>
> <element name="password" type="xsd:string" />
>
> <element name="options" type="impl:NamedValue"
> maxOccurs="unbounded" />
> </sequence>
> </complexType>
> </element>
> And the response like this:
>
>
>
> - <element name="loginResponse">
>
>
> - <complexType>
>
>
> - <sequence>
>
>
> <element name="loginReturn" type="impl:NamedValue"
> maxOccurs="unbounded" />
> </sequence>
> </complexType>
> </element>
> This seems OK so far, even though ArrayOfNamedValue is defined but it does
> not seem to be used.
>
> This is exactly what I want to pass back and forth:
>
> NamedValue object:
> name: "dummy-2"
> value: NamedValue[]
>
> But it doesn't look like that's what I get on the server side, it seems that
> I get:
>
> NamedValue object:
> name: "dummy-2"
> value: NamedValue <--- Note no array [] here
>
> And the value of the red NamedValue is the last value sent, in my example
> below it's 314
>
> How would Axis know that it needs to convert the request below into a
> NamedValue[] on the server side? What if the sequence of values would not be
> homogeneus, would it convert it into an Object[] instead (if it worked at
> all)?
>
> <login xmlns="http://some/namespace";>
> <username>tim</username>
> <password>tim</password>
> <options>
> <name>dummy1</name>
> <value xsi:type="xsd:string">dummy_val1</value>
> </options>
> <options>
> <name>dummy2</name>
> <value xsi:type="ns1:NamedValue" xmlns:ns1="http://some/namespace";>
> <name>dummy2-1</name>
> <value xsi:type="xsd:string">val2-1</value>
> </value>
> <value xsi:type="ns2:NamedValue" xmlns:ns2="http://some/namespace";>
> <name>dummy2-2</name>
> <value xsi:type="xsd:int">314</value>
> </value>
> </options>
> </login>
>
> The part in red above does not seem to map to this:
> NamedValue object:
> name: "dummy-2"
> value: NamedValue[]
>
> This confuses me greatly, I'm not sure whether what I'm trying to do is
> allowed/supported over wrapped doc/literal at all, if it is not I should
> then switch back to rpc/encoded or it's just a bug in Axis.
>
> Btw, .NET doesn't seem to be able to convert the red fragment above into a
> NamedValue[] so I suspect the message generated by Axis is not correct when
> it comes to arrays inside arrays.
>
> Thanks.
>
> Tim
>
>
> Anne Thomas Manes wrote:
> Tim,
>
> How did you define the array of NamedValue in the schema (where did
> the <options> tag come from)?
>
> It looks to me as if the message corresponds to the request. Because
> you're using <xsd:anytype>, the value element may contain an array of
> the NameValue. Hence the dummy-2 object is defined thus:
>
> NamedValue object:
> name: "dummy-2"
> value: NamedValue[]
>
> Is this not what you want?
>
> Anne
>
> On 4/13/05, Tim K. (Gmane) <[EMAIL PROTECTED]> wrote:
>
>
> Hello,
>
> If someone could please help me with this issue I would appreciate it. I
> recently converted my web service from rpc/encoded to wrapped
> document/literal so that it works with .NET (before the change everything
> still worked even with .NET, except for problems with custom
> exceptions/faults).
>
> I have a data type defined as:
>
> <complexType name="NamedValue">
> <sequence>
> <element name="name" nillable="true" type="xsd:string" />
> <element name="value" nillable="true" type="xsd:anyType" />
> </sequence>
> </complexType>
>
> I need the value to be xsd:anyType so that I can send arrays of NamedValue
> with various types for value.
>
> On the server side I have a method defined as:
>
> NamedValue[] login(String username, String password, NamedValue[] options)
>
> If I create on the client side an options array that looks like this:
>
> NamedValue[] options = new NamedValue[]
> {
> new NamedValue("dummy1", "dummy_val1"),
>
> new NamedValue("dummy2",
> new NamedValue[]
> {
> new NamedValue("dummy2-1", "val2-1"),
> new NamedValue("dummy2-2", new Integer(314))
> })
> };
>
> Note the array inside the array above.
>
> A request to the server from an Axis client (latest CVS version of 1.2RC3)
> looks like this:
>
> <login xmlns="http://some/namespace";>
> <username>tim</username>
> <password>tim</password>
> <options>
> <name>dummy1</name>
> <value xsi:type="xsd:string">dummy_val1</value>
> </options>
> <options>
> <name>dummy2</name>
> <value xsi:type="ns1:NamedValue" xmlns:ns1="http://some/namespace";>
> <name>dummy2-1</name>
> <value xsi:type="xsd:string">val2-1</value>
> </value>
> <value xsi:type="ns2:NamedValue" xmlns:ns2="http://some/namespace";>
> <name>dummy2-2</name>
> <value xsi:type="xsd:int">314</value>
> </value>
> </options>
> </login>
>
> The encoding of the array inside the array does not seem right to me, the
> server ends up with a NamedValue value for "dummy2" instead of a
> NamedValue[].
>
> If the message is not correct, what should it look like? I assume it would
> need a wrapper, something like ArrayOfNamedValue ... but then this type
> would have to be defined in the schema (which in this case it is), but what
> about say ArrayOfLong, ArrayOfString, etc. which are not defined in the
> schema and could be sent at runtime because the value type is xsd:anyType?
>
> Is this supposed to work the way I want it with wrapped doc/literal? It
> worked fine before with rpc/encoded.
>
> If so, is this a bug in Axis? Any available patches for it?
>
> Thank you for any help you may provide.
> --
> Tim
>
>
>
>
>
>
>
>
>
>
