254897 - SOAP encoding problem with complex structure and anyType

Status: RESOLVED DUPLICATE of bug 243904

(Core Graveyard :: Web Services, defect)

Description

[Cédric Chantepie]

User-Agent:       Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.7) Gecko/20040725 Firefox/0.9.1+
Build Identifier: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.7) Gecko/20040725 Firefox/0.9.1+

When using SOAP with complex structure encoder does seem make proper SOAP response.

Reproducible: Always
Steps to Reproduce:
1.
2.
3.

Actual Results:  
<enc:anyType>

Expected Results:  
<ns1:myComplexType>

Comment 1

[Cédric Chantepie]

Attachment: Patch for the trunk

Comment 2

[Cédric Chantepie]

We may need to allow to provide JS SOAP encoder impl. so that complex js object
would be encoded as user want

Comment 3

[Harshal Pradhan]

Great! This is something that I have seen also. I have been meaning to
investigate. It's one of the biggest issues that we have with asmx services
also. Thanks for doing this, I'll try to review and test this in the next few days.

(In reply to comment #2)
> We may need to allow to provide JS SOAP encoder impl. so that complex js object
> would be encoded as user want

Isn't this acutally possible if you just use the SOAP API? Its only the
wsdlproxy stuff that doesn't expose that capability outside. No?

Comment 4

[Cédric Chantepie]

> Isn't this acutally possible if you just use the SOAP API? Its only the
> wsdlproxy stuff that doesn't expose that capability outside. No?

How would you do that ?

Comment 5

[Harshal Pradhan]

Comment on attachment 155584 [details] [diff] [review]
Patch for the trunk

I can't think of anything that this might break.
But this stuff is very fragile so please test carefully
before checkin.

r=me if you fix the following nits. 


>   nsresult rc;
>-  PRBool needType = PR_FALSE;
>+  PRBool needType = PR_TRUE;

Move declaration of needType below, closer to where it is first needed.

[snip]

>     // find the base type and use the name of that as the element name.
>     ns = gSOAPStrings->kSOAPEncURI;
>     nsAutoString currentURI = ns;
>-    nsCOMPtr<nsISchemaType> currentType = aSchemaType;
>+    nsCOMPtr<nsISchemaType> currentType = aSchemaType;    

Don't add unnecessary whitespace noise.

>-    rc = aEncoding->GetExternalSchemaURI(gSOAPStrings->kSOAPEncURI, ns);
>+
>+    if (!typeNS.IsEmpty()) {
>+      ns.Truncate();
>+      ns.SetIsVoid(true);
>+
>+      rc = NS_OK;
>+    }
>+    else {
>+      rc = aEncoding->GetExternalSchemaURI(gSOAPStrings->kSOAPEncURI, ns);
>+    }

SetIsVoid() is not needed as far as I can see. Get rid of it
if its not doing anything useful. Truncate() is enough.

Comment 6

[Cédric Chantepie]

Attachment: Fixed patch [trunk]

Comment 7

[Johnny Stenback (:jst)]

Comment on attachment 158237 [details] [diff] [review]
Fixed patch [trunk]

sr=jst

Comment 8

[Mike Kaply [:mkaply]]

Comment on attachment 158237 [details] [diff] [review]
Fixed patch [trunk]

This is a Gecko change that would cause a difference between 1.7 and Firefox.

Comment 9

[Cédric Chantepie]

*** This bug has been marked as a duplicate of 243904 ***