W3C

- DRAFT -

XML Processing Model WG

27 Jan 2016

See also: IRC log

Attendees

Present
Norm, Murray, Alex, Jim
Regrets
Henry
Chair
Norm
Scribe
Norm

Contents


Alex: I don't agree with Henry (from email); I don't want to go back to first principals. We've done all the requirements analysis.
... I want to build something useful, not recast the requirements yet again.
... We've done lots of analysis.

Murray: I acknowledge that you've been doing this for a long time. But it's only been a few months since the new ideas came forward as the beginning of a proposal.
... It feels like things aren't being explained very well. I think we should talk today, even though I said I agree with Henry.

Alex: I feel like we have requirements; they're a guide. We need to go back to them to make sure we do the right thing in the spec.
... That was always true; this isn't XProc 2.0, this is a proposal for a different direction.
... There's no consensus that this has any standing at all.

Murray: Let's move beyond the meta-conversation and have the conversation.

Norm: I wanted to talk about the new syntax proposals.

Alex: I think we should talk about the append operator, >>
... We also need to talk about the upcoming face-to-face.
... If we're going to polish this as a proposal, we need to make it read like something mortals could understand.

Alex reviews the latest version of the document.

Alex: What Henry identified was that $1 on the left hand side has a different meaning than when it was on the right hands side.
... We need a way to have a positional output.
... The farthest right-hand side of things, when there's an output that needs to go someplace, we assign it to a variable..
... That's not the same as "send the output of the previous to the output of the next". It's a different thing.
... I thought we needed a different operator for that.
... Jim and I chatted about it and came up with ">>".
... There happens to be a single Unicode character for it.
... So now there's nothing to do with output in the first section.
... I did add square brackets in the 'Step Chains' section.
... It's an ordered list of sorts, it's an array kind of thing.

Alex reviews Appending Chains and Outputs

Alex: The identity becomes $in >> $out.
... We can use >> "filename.xml" for storing to that URI.

Murray: Are you appending to that file, or replacing it?

Alex: Appending if it exists.
... If you do it more than once, we have to define what it means.
... We have to define the semantics for what "append to a URI" means

Murray: It's really confusing if you say this is what it means and then say that you have to define what it means.

Some discussion of the order of outputs in >> $included. Order is implementation determined.

Some discussion of the use of $1 in XPath expressions

Alex: You're going to have to do static analysis on the expressions. There just isn't going to be any way around that.

Jim: Do we have a default context within the curly braces?
... Could it be the "." instead of $1?

Alex: We're going to have to define the context.
... In a block expression, you're going to have to have a way of talking about positional inputs and ouputs.

Murray: I understand about positional inputs. There's more than one stream coming into this process.
... Or more than one command line argument.

Norm attempts to explain xslt() -> { $1, $2 }

Murray: When I see a '$', I think of a variable. But we're using them to be a variable representation of a port.
... When I suggested @, I was suggesting @ instead of $.
... If you're going to create a reference to a port then any subsequent uses of that port would use @.
... So you'd write to @out, >> @out, and then later you'd refer to @out.

Alex: We could have a different syntax for port references. That would be more consistent with @1, @2, etc.
... We have scoping questions that go along with that.
... You have scoping rules for port references and for variable references.
... My want had been to make the rules on the input side be the same. We should be able to use the output port assigned to some variable as a variable reference.
... There's an immediate correlation between the variable output port and the variable in the expression.
... I need a way to talk about the input and make an expression against it.
... If we use a different syntax ,then we don't get that for free. We have to have some way to convert a port name into a sequence of things.
... Some variables point to ports that have a conversion syntax within XPath.

Murray: I just want to observe that I understood the arguments about using variables. If we have variables and port names, you wouldn't have to use the port names.
... You could use variables everywhere. But you could write it out "full-hand".
... Part of the message I'm trying to give here is that I want to see the long version before the contraction.

<jfuller> FWIW I am fine with @ on left hand side .. I think it clearly delineates ports vs options

Meeting planning

<jfuller> XML Prague

<jfuller> http://www.xmlprague.cz/day1-2016/

Norm: I propose three half days: half a day to work on the proposal, half a day to prep for the public meeting, half a day for that meeting
... An introduction, a presentation of the new syntax, and discussion

Jim: I think we need an answer for XML syntax for this language.
... I don't know what that would be.
... One question will be: what do we do now that we don't have an XML syntax.

Alex: We only get a few vocal people.

Any other business?

Jim: For the f2f meeting, for people who aren't attending, is there going to be any attempt to have telepresence?

Adjourned

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2016/02/02 00:31:15 $