445285 - An instance with an empty default namespace inherits the namespace of the parent document.

Status: RESOLVED FIXED

(Core Graveyard :: XForms, defect)

Description

[Swithun Crowe]

User-Agent:       Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.8.1.14) Gecko/20080416 BonEcho/2.0.0.14
Build Identifier: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.8.1.14) Gecko/20080416 BonEcho/2.0.0.14

The XHTML document has a default XHTML namespace. An inline instance has a non-default namespace and an empty default one. On submission, the instance's default namespace has become the XHTML one.

Reproducible: Always

Steps to Reproduce:
1. Load form like http://swithun.servebeer.com/namespace.xhtml
2. Submit
3. Check output
Actual Results:  
The submitted instance is:

<?xml version="1.0" encoding="UTF-8"?>
<foo:Foo xmlns:foo="http://www.example.com/#foo" xmlns="http://www.w3.org/1999/xhtml" 
xmlns:xf="http://www.w3.org/2002/xforms" xmlns:ev="http://www.w3.org/2001/xml-events" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"/>

Expected Results:  
Ideally, you would get:

<?xml version="1.0" encoding="UTF-8"?>
<foo:Foo xmlns:foo="http://www.example.com/#foo"/> 

Formsplayer produces the desired result. From what the W3C says about inheriting/overriding default namespaces (http://www.w3.org/TR/REC-xml-names/#defaulting), I would say that the XForms plugin isn't handling this correctly.

Comment 1

[Nick Van den Bleeken]

Someone (Swithun Crowe) raised this issue also on the www-forms mailinglist of w3c, I informed him that it is a bug in the Mozilla XForms implementation.

http://lists.w3.org/Archives/Public/www-forms/2008Jul/0001.html

Comment 2

[Leigh L. Klotz, Jr.]

Possibly related to #330557 

Comment 3

[Leigh L. Klotz, Jr.]

I believe the issue is serialization, not construction of the InfoSet.
The attached updated namespace2.xhtml shows that the the instance has child nodes of foo:Foo in the xmlns="" namespace.

Also, the attached example removes the Ubiquity script reference; it appeared to make no difference.

Comment 4

[Leigh L. Klotz, Jr.]

Attachment: Shows issue is not with namespace construction, but with serialization; removes Ubiquity script reference

Comment 5

[aaronr]

Not really related to bug 330557.  The fact that we are serializing by hand and possibly missing namespaces from some ancestor elements is what bug 330557 is about.  If we can figure out how to serialize namespaces through mozilla and eliminate the need for serializing by hand, then we can get rid of the code that contains this bug.

Comment 6

[aaronr]

Attachment: patch

added a parameter to AddNameSpaces to allow us to remember whether to ignore the default namespace as we search up our ancestor tree

Comment 7

[Olli Pettay [:smaug][[e-mail]]]

Comment on attachment 330356 [details] [diff] [review]
patch


>     nsAutoString XMLNSAttrValue;
>+    PRBool ignoreDefaultNS = PR_FALSE;
>     submDocElm->GetAttributeNS(kXMLNSNameSpaceURI, NS_LITERAL_STRING("xmlns"),
>                                XMLNSAttrValue);

Should we actually test HasAttributeNS? It is ok to have empty value.

> 
>     if (XMLNSAttrValue.IsEmpty() && (!prefixHash ||
>         !prefixHash->Get(NS_LITERAL_STRING("#default"), nsnull))) {
>       submDocElm->RemoveAttributeNS(kXMLNSNameSpaceURI,
>                                     NS_LITERAL_STRING("xmlns"));
>+      ignoreDefaultNS = PR_TRUE;
>     }

Comment 8

[aaronr]

On this thread, http://lists.w3.org/Archives/Public/www-forms/2008Jul/0001.html, the W3C said that it doesn't matter if we serialize an empty default namespace or not, but as of Friday morning things were leaning toward not serializing it as a preference.  More arguments have come up since then, though.

I'm fine with serializing the empty default namespace.  If you feel the same, then I'll just do it and we'll be done with it.

Comment 9

[Olli Pettay [:smaug][[e-mail]]]

I think the default namespace should be serialized.

Comment 10

[aaronr]

Attachment: patch2

Changed per Olli to export the default namespace.  No real consensus on the W3C side whether it should be serialized or not, either, so I guess if the author bothered to put it there, we should serialize it.

Comment 11

[aaronr]

Attachment: testcase w/ diff echo server

same testcase as before but using the xformstest.org echo server (which seems to always be up and the other one doesn't)

Comment 12

[aaronr]

checked into cvs trunk

Comment 13

[aaronr]

checked into 1.8 branch