Re: XML InfoSet and property value preservation

Wilfredo Sánchez Vega wrote:
>   Right, but the fact is that XML parsing libraries don't generally 
> preserve prefix names.  So I agree that those were bad ideas.

What do you mean by generally? DOM - as far as I can tell the only XML 
API with an W3C stamp on it - preserves them. So does SAX.

In fact, building a compliant XSLT processor on top of an API does 
require that.

>   So unless the XML spec is going to change to require parsers to 
> preserve prefix and we can require XML parsers to comply with the 

The XML spec requires that.

> updated XML spec, I don't think it's reasonable to expect a server to 
> preserve prefixes.  Doing so requires the server author to use an XML 

With all due respect, I think you need to check the facts.

> library that does more than the XML spec requires, which may mean they 
> have to write their own.  If you can't use a standard XML library, 
> you're obviating the main advantage of using XML at all.
> 
>   If you need to preserve prefixes, regardless of what your reasons are, 
> including your use of other (perhaps poorly thought out) W3C specs, you 
> are in effect expecting the server to treat your XML as text.  If that's 
> the case, then encode the property XML as text and be done with it.  I 
> don't really follow why this isn't considered a viable option here.

Wilfredo,

I think you're arguing based on a misunderstanding of XML. The XML spec 
itself doesn't say anything about the prefix in a name being ignorable 
(as a matter of fact, it doesn't speak of prefixes at all, because 
that's in the Namespaces spec). Nor does the XML Namespaces spec itself. 
Nor does DOM, XPath or XSLT. Or SAX (for an example of a non-W3C API).

Best regards, Julian

Received on Thursday, 17 November 2005 20:17:45 UTC