Re: Case where well formed ness could fly away

On 5/1/07, Alex Milowski <alex@milowski.org> wrote:
>
>
>
> On 5/1/07, Innovimax SARL <innovimax@gmail.com> wrote:
> >
> >
> > The point is : if on input there is sequence of well formed document,
> > is it possible to have non well formed document as output
> >
> > Count
> > Impossible
> >
> > Delete
> > Impossible
> >
> > Error
> > Impossible
> >
> > Escape Markup
> > Impossible
> >
> > Head
> > Impossible
> >
> > Identity
> > Impossible
> >
> > Insert
> > Not sure
>
>
> Since you are inserting a child into an existing element, this should be
> OK.
>
> Label element
> > Impossible
> >
> > Load
> > Impossible
> >
> > Matching documents
> > Impossible
> >
> > Namespace rename
> > You have to take care of Uniqueness of Attributes
> > example
> > <a ns1:a="2" ns2:a="2" xmlns:ns1="ns1" xmlns:ns2="ns2">..</a>
> > rename from="ns2" to="ns1"
>
>
> Namespace renames are by URI.  This shouldn't happen if the input document
> is well-formed already.
>

Sorry I use the same value for prefix and namespace URI , but I was speaking
of URI, please reread with that in mind

Rename
> > Can fail, but case are identified
> >
> > Replace
> > Impossible
> >
> > Set attributes
> > You have to take care of Uniqueness of Attributes
> > <a ns1:a="2" xmlns:ns1="ns">
> >
> > attributes : <root ns2:a="3" xmlns:ns2="ns">
>
>
> Attributes are merged and when they have the same name, the "attributes"
> port
> wins.
>

what is the same name ? local name or QName ? in the case I present what
would be the result ?

I hadn't intended namespace attributes to be handled by this step.
>

namespace attribute ? why are you talking about that ?

String Replace
> > Not sure
> >
> > Store
> > Impossible
> >
> > Tee
> > Impossible
> >
> > Tail
> > Impossible
> >
> > Unescape Markup
> > Can fail in many ways
>
>
> Yes, any parse error should cause the step to fail.  I should probably add
> that.
>

That would help

Unwrap
> > Can fail, one case is identified, but what about removing the node
> > which contains namespace declarations ?
>
>
> Namespace declaration fixup will be necessary  during serialization for
> the results of many steps--not just unwrap.
>

I'm sorry, I **strongly** disagree. What is you definition of serialization
? The namespace fixup should happen even when the flow will never be
serialized. It's a DOM or SAX or StAX issue.

As far, as I know, I will still have to be able to make a select on the
resulting document


As such, that's a serialization
> problem.  Between steps, the namespaces and local names are known
> and implementations that do not serialize between steps will not have
> to worry about that.
>


I think you miss the question

<a xmlns:ns2="ns1URI">
  <b xmlns:ns2="ns2URI">
    <ns2:c/>
    <ns2:c/>
    <ns2:c/>
    <ns2:c/>
  </b>
</a>

I want to unwrap "b", will i have

<a xmlns:ns2="ns1URI">
    <ns2:c/>
    <ns2:c/>
    <ns2:c/>
</a>

or

<a xmlns:ns2="ns1URI">
  <ns2:c xmlns:ns2="ns2URI"/>
  <ns2:c xmlns:ns2="ns2URI"/>
  <ns2:c xmlns:ns2="ns2URI"/>
</a>

or something else ?

We should probably make sure we have a general statement about
> this somewhere in the spec.
>

I agree, but I fear the problem is specific to unwrap in that case

Wrap
> > Can fail with a non well formed name
>
>
> Yes.  I should add something about the qname value of the option
> resolving properly.
>

Mohamed



-- 
Innovimax SARL
Consulting, Training & XML Development
9, impasse des Orteaux
75020 Paris
Tel : +33 8 72 475787
Fax : +33 1 4356 1746
http://www.innovimax.fr
RCS Paris 488.018.631
SARL au capital de 10.000 €