Re: [xml] Ability to make parser ignore the XML_NS_ERR_UNDEFINED_NAMESPACE error

[Greg Shtilman <gmatroskin yahoo com>]

Daniel,

Thanks for youur quick response.  And thanks for the
suggestion on filtering errors by domain.  I havent
explored the new structured errors to the full extent
yet, I'll look into it.  

I did notice that libxml continues to parse the tree,
but I wasn't sure whether we can count on this
behaviour staying in the future releases.  Also the
comment before xmlNsError declaration  
(Handle a fatal parser error, i.e. violating
Well-Formedness constraints) mislead me shomewhat.

What you are suggesting will work for us except for
one little detail:  When the tree parsed in such a
way, is dumped, namespaces with missing declarations
are stripped from the output.  Any way of going around
this?  

Thanks,

Greg

--- Daniel Veillard <veillard redhat com> wrote:
> On Tue, Mar 09, 2004 at 04:43:20PM -0800, Greg
> Shtilman wrote:
> > Hi,
> >
> > I am trying to find out why a missing namespace
> > declaration was made a FATAL error in 2.6.x?  We
> use
> > libxml to parse 'snippets' of XML that are missing
> the
> > namespace declarations by design.  We used to set
> the
> > ignore warnings flag in the older libxml, but are
> > considering an upgrade to 2.6.6 now, and
> discovering
> > that the only way to make libxml 'happy' is to
> > surround the document with a dummy root node that
> has
> > the namespace declarations prior to parsing, and
> > removing them when dumping.  Would it be possible
> to
> > provide a flag, or some other means (such as the
> > ability to predefine the namespaces) to make these
> > warnings ignorable in such a case?
>
> paphio:~/XML -> cat tst2.xml
> <doc s:b="foo"/>
> paphio:~/XML -> xmllint tst2.xml
> tst2.xml:1: namespace error : Namespace prefix s for
> b on doc is not defined
> <doc s:b="foo"/>
>               ^
> <?xml version="1.0"?>
> <doc b="foo"/>
> paphio:~/XML ->
>
>   it is a namespace error, not a well-formedness
> error. libxml2
> still continues to parse the document and produce a
> tree in that case.
> You can filter errors by domain and catch namespace
> errors. See 
> <libxml/xmlerror.h> for a the associated APIs and
> definitions.
> It's the first time someone complains about the new
> parser behaviour,
> in itself it's an interesting point.
>
> Daniel
>
> -- 
> Daniel Veillard      | Red Hat Network
> https://rhn.redhat.com/
> veillard redhat com  | libxml GNOME XML XSLT toolkit
>  http://xmlsoft.org/
> http://veillard.com/ | Rpmfind RPM search engine
http://rpmfind.net/


__________________________________
Do you Yahoo!?
Yahoo! Search - Find what you?re looking for faster
http://search.yahoo.com