W3C home > Mailing lists > Public > xproc-dev@w3.org > December 2008

Re: Entity Expansion

From: Norman Walsh <ndw@nwalsh.com>
Date: Sat, 27 Dec 2008 11:16:38 -0500
To: XProc Dev <xproc-dev@w3.org>
Message-ID: <m2d4fdwq3d.fsf@nwalsh.com>
"David A. Lee" <dlee@calldei.com> writes:

FYI: this is an area that the WG is actively discussing, so it's
possible that some of what I say below will change.

> What I mean is this.  In an entity-expanding implementation, then I
> would presume that p:identity WOULD expand entities, and there's
> nothing in the specs to say it would have
> to keep track of the base URI's for the elements it expanded.

Ah. I think you're mistaken there. The base URI infoset property
records the "URI used to retrieve the entity" per section 5.1 of
RFC 2396.

And XProc says, in 2.4.1:

  Except where the semantics of a step explicitly require changes,
  processors are required to preserve the information in the documents
  and fragments they manipulate. In particular, the information
  corresponding to the [Infoset] properties [attributes], [base URI],
  [children], [local name], [namespace name], [normalized value],
  [owner], and [parent] must be preserved.

So I expect the base URI to be preserved through p:identity steps.

> This would then make a test like add-xml-base-001 fail because the
> p:identity would have lost where every element came from ... unless it
> was a requirement that any expanded entity kept the  base URI through
> the entire pipeline which would be yet a more complex requirement then
> I think you (or I) want to impose. 

Every element (and PI, though that's less commonly a concern) has a
base URI property. Once set, it should be preserved.

> I dont think it could be done
> with adding xml:base attributes everywhere because it would break the
> rest of the specs that assume, unless otherwise noted, that no new
> attributes are added.
> Thus the only way an implementation could preserve the base URI's
> would be some 'secret back door channel' that preserves node identity
> and base URI's throughout the pipeline.  Yuck.

You don't need a secret back door, most APIs provide the base URI of
an element, you just need to make sure that gets preserved across
steps.

> So in conclusion, What I think is true ...
> 1) The specs dont say anything one way or another about entity
> expansion, although there's the implicit assumption that they probably
> are expanded at any conceviable step. 

Yes.

> And presumably,
> since its not stated, dont have to keep track of where the expansion
> came from (base uri).

No.

> 2) add-xml-base-001 is actually not testable as-is because there is no
> requirement that implementations actually preserve the base URI when
> they expand entities, or do or do not expand entities.

No.

> I certianly think the "intent" of add-xml-base-001 is correct, and I
> will try to make my implementation match it, but I dont think its
> testable, that is, I dont think it should strictly be considered a
> failure if the step didnt expand entities, or somehow lost where the
> expanded node came from.

If your implementation doesn't expand entities, I think you could
argue that you pass that test if the results you give are consistent
with unexpanded entities.

I don't agree that you're allowed to forget the base URI of elements.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | Next to knowing when to seize an
http://nwalsh.com/            | opportunity, the most important thing
                              | in life is to know when to forego an
                              | advantage.--Benjamin Disraeli

Received on Saturday, 27 December 2008 16:17:18 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Saturday, 27 December 2008 16:17:18 GMT