Re: Intent of ER-XML

On 27/02/2012 01:46, David Lee wrote:
> I could see several opportunities. E.g. should XInclude or external
> entities also invoke the XML-ER processor ? Should that be defined by
> the spec or left up to implementers ?


well definitely still to be decided. I've suggested that the entire
doctype be essentially skipped (so external entities wouldn't get
expanded) but various other things are possible and may make more sense.....


> A processor cannot simultaneously be a XML 1.x AND a XML-ER
> processor at the same time. Which means they are definitively
> different things. An XML-ER processor ("parser"?) cannot be an XML
> 1.x parser ... at least at the same time.   An implementation could
> provide both but not simultaneously.

May as well call the xml-er parser a "processor" that's what the xml
spec calls xml parsers (even if no one else does:-)

>
> Hence the intent or meaning of "drop in" is hard to define ...

one definition would be that it should be possible to substitute an
xml-er processor for an xml processor using the same API (whether that
be DOM or SAX or whatever) and that any application using that processor
will be able to process the input as if it were well formed XML.

I think that's a requirement (not met by the current draft as it doesn't
enforce element names, but I'm not worried about that at this stage)

One could refine that by placing constraints on what the well formed xml
document looks like. The strongest statement I think would be that
any namespace well formed xml document produced the same tree using xml
1.1 or xml-er. but I think that might be too strong and we might not
want to enforce entity handling in which case we could say that any
namespace well formed document with no doctype declaration produces the
same tree using xml 1.1 and xml-er.


> I have little problems with the spec as-is as a start.  My discussion
> is more a step-back to discuss how to go forward and to do that we
> need an agreement on the "philosophy"... that is what are we trying
> to describe ?   Without that being clear I can't even analyze the
> spec in more detail to have an opinion.

yes, agreed.

David

Received on Monday, 27 February 2012 02:10:51 UTC