Accept this agenda?

-> http://www.w3.org/XML/XProc/2016/02/24-agenda

Accepted.

Accept minutes from the previous meeting?

-> http://www.w3.org/XML/XProc/2016/02/10-minutes

-> http://www.w3.org/XML/XProc/2016/02/11-minutes

Jim: Liam minuted the public meeting and we don't have those notes yet.

Alex: Aren't they in the IRC log?
... I don't have the diagram that one of the participants drew.

Jim: I think Liam has it.

Alex: Ok, we need a copy of that.

Jim: Ok, those are the minutes I was thinking of.

Norm: So approve?

Jim: Ok.

Accepted.

Next meeting, 2 Mar 2016

No regrets heard

Reports from the face-to-face

Norm: Anyone have stories to tell not in the minutes?

None heard

Community group status/update

Jim: It's good to get people into a community (with a small "c") and I wanted to help as much as I can.
... There's a little admin involved. At XML Prague, Nic and Ari said they'd be willing to co-chair. We discussed
... what items the community group could do. We came up with three: shepherding the XML syntax, because there are
... folks interested in doing active work there. I've had meetings with some of these folks. There's going to be an announcement
... that we want to have coincide with the official announcement.
... We're working on a transform for that.
... I also thought that a new name might be a good idea; Nic is going to setup an issue tracker.
... They seem like a really good fit.
... I'm also coordinating to meet with them a few times this year.

Norm: I propose that we plan to invite them to attend this call periodically to give feedback.

<ht> +1

Jim: I'm hoping to get things aligned between what we could use help on and what their agendas are. We'll take it step-by-step.
... I'll give an update next time.

Murray: There is a liason process within W3C, right? They can communicate with us, right?

Norm: Yes.

Autumn workshop

Jim: The timing didn't work for what we initially discussed at XML Prague. All the efforts are now poured into doing an XProc workshop in Amsterdam at CWI with Steven Pemberton.
... No idea what the attendance will be, but I think we can guarantee enough to justify the meeting.
... Should I keep working on that.

Norm: YES PLEASE
... I still think a more general "pipeline" or "data flow" workshop is a great idea, maybe for spring of 2017.

Jim: I'll work on getting some cross-pollination going for that conference.

Murray: It occurs to me that we could call it "pipefitters" or "XML pipefitters"

Henry: What about a face-to-face at Murray's in the summer?

Norm: Henry, we were waiting on dates from you. :-)

Henry: What would work for me would be early the week of 20 June.

Norm: Ok, Murray does that work for you?

Murray: It seems plausible. Let me double check.

Jim: I can't commit.

Alex: I don't know either; summer is difficult.

Update the repo

Norm: I'd like to clear the decks in the spec repo.

Henry: I think moving it all to an "archive" directory or something would be fine.

Murray: How about renaming it

Norm: Well, duh. Yes, thank you, that's a good idea.

Murray: Going back to the community group topic for a moment, what "xflowml" or "dataflowml"?

<alexmilowski> The letter ‘X’ shall not be used … ;)

<alexmilowski> Don’t do a “s/X/J/g” in IRC ...

Murray: I like "flowml" because I like the way the "ml" sounds.

<ht> Murray: As in "Flow 'em all"

<alexmilowski> Not a joke to some …

Norm: I don't hear anyone objecting to my jiggering with the spec repos, so Imma gonna.

Approved.

Demo of Jim's efforts

Jim: It's not anywhere public at the moment, but I'll be making it available at some point when it's a little more mature.

<ht> Not a syntax designed for hand-authoring -- yes

<alexmilowski> Annotations!

Jim's demo is a thing of beauty.

Murray: Thanks, Jim, that's very cool.

Data literals

-> https://lists.w3.org/Archives/Public/public-xml-processing-model-wg/2016Feb/0021.html

Norm: What about AVTs?

Alex: AVTs are an artifact of XML. I don't think we need them, we should go for simple.
... We should have a JSON templating step and an XML templating step and a everything else templating step.
... Pick your poison. Get a step that provides it.
... Mustache templates, for example.

Norm: Ok, I guess we can run with that for a while.

Henry: The problem I have is that it feels like the right decomposition is that we ought to have tools for constructing string literals with templating.
... But it doesn't do the right thing because the templating step needs to know about the data model.
... If it were XProc1, it would be fine, you could build the string and parse it.

<liam> [I thought I sent the log from te public meeting to the public list, sorry!]

Henry: But that doesn't work in XProc2 because variables can have structure.
... You can't use XML with the JSON templating, or JSON with the XML templating

Alex: There's a story that's been well worked out that given some set of values in XQuery there's a template you can construct in XML and only XML
... Everything else in the universe is all over the planet.
... On the left hand side is random syntax that the step expects.

<liam> [ unedited log from publc meeting: https://www.w3.org/2016/02/11-xproc-irc -- I'll try & tidy it up today, modulo the raging ice storm outside taking away our electricity ]

[ Check my minutes of 11 Feb, I might have found it; Jim was initially confued ]

<liam> [ ok ]

Henry: There's no step here, there's no JSON specific step.

Alex: What I'm saying is that templating is out of scope.

Henry: I'm not disagreeing with you, I just don't understand your proposal.

Alex: What I'm saying is that if we have a convenient syntax for literals, we can stop there.

Norm attempts to explain: the data literal syntax is just about string literals.

data "application/json" '{"fruit": "orange"}' -> x:mustache($options) >> $output

Henry: There's nothing in the data literal that looks like any kind of JSON templating syntax.

Alex: The step has an expected media type and it gets what it gets, it just doesn't matter.

data "application/json" '{"fruit": "orange"}' -> x:change-oranges-to-apples($options) >> $output

Henry: I understand how the language works, but I don't understand how that can work for templating json.

Norm: I'm confused.

Henry: I remain to be convinced that the input to a JSON templating step is actual valid JSON. And if it takes strings, that works. But I'm skeptical that something that takes application/json as input can be a flexible JSON templating solution.

<alexmilowski> https://mustache.github.io/mustache.5.html

Norm: Alas, we're out of time.

Alex: That link is just a text templating thing. It takes text and outputs text.

Any other business?

None heard. Adjourned.