Re: [xml-dev] Versioning MicroXML (Was: MicroXML)

From
Liam R E Quin <>
To
Pete Cordell <>
Date
2010-12-14T17:33:44Z
ID
<>
Thread
Re: [xml-dev] Versioning MicroXML (Was: MicroXML)
On Tue, 2010-12-14 at 12:13 +0000, Pete Cordell wrote:

> Taking ideas from various internet protocol negotiation schemes, can we 
> include a uxmldecl that includes a "require" attribute that lists features 
> the uXML parser must support.
> 
> For example:
> 
> <?uxml require="ns pi"?>

As soon as you do that your spec no longer defines a single language,
but 2^n languages, where n is the number of things you can write.
It really hurts interoperability.

Language negotiation lets the server choose the single preferred value.
So you only need n variants, not 2^n

You did go on to say,
> 
> BTW - I think its important to agree that we are only looking for 3 or 4 
> widely implemented features, rather than 3 or 4 million!  If we have too 
> many then we lose interoperability.  Things analogous to XML namespaces and 
> xml:id that were added to XML after the initial release of XML are the sorts 
> of things we should be looking for.

so, so far you've listed 3, and I can easily see a dozen more happening.
comments
lax error handling
processing instructions
namespaces, namespace path, external namespace definition files
xml:base
xml:id
xinclude
microschema, microentitydefinitions... :)
mixed content and xml:space
and so on and so forth

Better to say, either use microxml or xml.

XML succeeded largely because of the XML Promise -- every XML processor
is licensed by the XML Specification to process any XML document.

I *do* think there would be mileage in a hook for later extensions,
e.g. reserving <µxml>...

Liam

-- 
Liam Quin - XML Activity Lead, W3C, http://www.w3.org/People/Quin/
Pictures from old books: http://fromoldbooks.org/
Ankh: irc.sorcery.net irc.gnome.org www.advogato.org