See also: IRC log
-> http://www.w3.org/XML/XProc/2015/01/07-agenda
Accepted.
-> http://www.w3.org/XML/XProc/2014/12/10-minutes
Accepted.
Proposed: 14 January 2015
No regrets heard.
Alex reports no progress on his items.
Jim reports no progress on his items.
Norm reviews the idea of having "XML stub documents" to represent non-XML documents in the pipeline.
Jim: What about the base URI?
<alexmilowski> http://www.w3.org/TR/xpath-datamodel/#DocumentNode
Norm: It is what it is.
Alex: I think maybe we could use
a stub "XDM Document" to represent binaries.
... If you access the children you get a base64 encoded
representation or a JSON blob or some such.
... That would leave open the possibility of unification later;
you'd still have the concept of a document node but you'd have
other kinds of children.
... JSON could be represented with semi-structured data.
... But that's the future, all we want today is properties,
right?
Norm: Yes.
Jim: Is this possibly a little
bit too much sausage making. From a first-principles point of
view, we're allowing non-XML documents to flow.
... From a pipeline author's point of view, they're just going
to get a binary document.
... To me the simplest thing is a binary flowing through.
<alexmilowski> What is "this proposal" ?
Jim: Are we saying that we're passing references around? This is more implementation level detail.
Alex: I'm not sure what you mean.
The proposal Norm outlined is a variation of something I
proposed a long time ago that you can find in the
archives.
... What I'm saying today on the call is that we can do it with
XDM document nodes.
... I think there's room in the current XDM to create some kind
of Document node that has "on demand" content.
... Of course, we run the risk that someone will say we have to
fix these things in the XDM.
... I'd like it to be possible for vendors to use extensions to
the XDM to represent things like JSON if they wanted to.
Norm waffles a bit about how implementors will react to this.
<ht> Not clear to me why the string value isn't always empty
Alex: What about plain text?
Norm: I think those can just be Document nodes with a single Text node child. That's what XSLT does.
Alex: There are two issues here: there's a story about random media types that you have and how they map into a Document node and then there's a possibly not-normative description of what you do with text/plain.
Henry: I'm on a fine red line
here. These are not XML documents, these are not XDM nodes,
there is no string content, it's always empty. There's an
accessor to get at the content.
... It's a recipe for confusion if there's any sort of
flexibility here at all.
Some discussion of text/plain documents.
Alex: For semi-structured data like JSON, I want it to be possible for implementors to extend the XDM in ways to access JSON.
Norm gives a little background about extending the XDM to JSON.
Henry: Fine, we can do that in
XProc 2.1. I think we need a complete and straightforward story
about what these documents contain.
... Extending the XDM is a new version of the spec.
Alex: I understand the concerns, we can make an issue about that.
Jim: To help us with the story, I
think we're jumping around a bit in use cases. Would it be good
to create a set of pipelines that demonstrate some of these use
cases.
... how far do we think this is going to go.
... Are we going to run p:viewport over binary documents?
Alex: I think the simplest use case is just the ability to get the base URI.
Norm: No, p:viewport only operates on XML, lots of steps only process XML.
Some discussion of ZIP file manipulation.
Some discussion of filtering p:for-each based on media type.
Jim: There's a JSON use case,
packaged up stuff (EPUB, zips), image processing use
case.
... Anything else?
<scribe> ACTION: A-262-01 Jim to attempt to describe these use cases, write demonstration pipelines. [recorded in http://www.w3.org/2015/01/07-xproc-minutes.html#action01]
Alex: I think we should be explicit about the text/plain case.
Norm: I think the next step is to consider what violence it does to the spec to attempt to describe binaries as having stub XDM Document nodes.
<jfuller> xml prague
<jfuller> http://www.xmlprague.cz/sessions2015/
Alex: I think we should have some
working pipelines that show off AVTs and stuff.
... Maybe you (Norm) can send an email outlining what you think
we should cover.
Adjourned.