XML Processing Model WG

Meeting 192, 21 Apr 2011


See also: IRC log


Norm, Paul, Henry, Alex, Mohamed


Accept this agenda?

-> http://www.w3.org/XML/XProc/2011/04/21-agenda.html

Norm: I think we'll take the comments in backwards order.

No objections.

Accept minutes from the previous meeting?

-> http://www.w3.org/XML/XProc/2011/04/14-minutes.html


Next meeting: telcon, 5 May 2011?

No regrets heard.

Norm will be unavailable 12 and 19 May

Browsers and profiles

-> http://lists.w3.org/Archives/Public/public-xml-processing-model-wg/2011Apr/0014.html

Norm summarizes how he got there.

Henry: I don't have any problem with that. I think the suggestion is basically good.
... I wish the world was such that the way I like to work would be better supported, but that doesn't seem likely.
... What worries me about the Firefox decision is that it may be the sharp end of a long wedge to drive XML processing out of the browser

Norm: Yes, I think the browser vendors are likely to go that way.

Henry: I replied suggesting that looking at the standalone setting might be useful.
... But the default for standalone is "no" so that's not really much help.

Some discussion of what the consequences of standalone settings would be.

Henry: What would be ideal would be if an explicit standalone=no was the only way to request retrieval of external declarations.

But that ain't the way it is.

Norm: I think the other part of my proposal that might be worth discussing is that I threw out XInclude

Henry: I thought Henri Sivonen indicated that he had a requirement to support XInclude which surprised me.

Norm: I didn't read the message that way, but ok...
... I like having three profiles better than four and if we were going to do XInclude I think we'd need a fourth and I didn't want a fourth.

Alex: Is that really necessary?
... I think there's a difference between XInclude and external subset fetching.
... External subsets are an artifact of identifying document types and they get fetched over and over again. But XInclude is more like images or links, they're explicit requests by the author.
... You don't want to go do the fetch, don't put the XInclude in.

Norm: Yes, I agree Xinclude is more like images and scripts, but it's more complicated. It's XML. It has fallback, it's recursive, etc. I don't think the browser vendors will do it.

Alex: We could add XInclude and send it out as another Last Call and see what happens.

Henry: I'd like to do a little negotiation in advance.

Norm expresses some reluctance to have another draft with a recommended profile that isn't what the browsers do.

Some discussion of the consequences of doing XInclude in JavaScript. The rendering won't be entirely progressive.

Alex: What about base URI fixup?


scribe: You might be able to do it with the xml:base attribute, but you can't specify the base URI directly when you create a node in JavaScript.

<scribe> ACTION: Alex to investigate the possibility of doing XInclude in a conformant way only with JavaScript. [recorded in http://www.w3.org/2011/04/21-xproc-minutes.html#action01]

Alex: There's a lot of animosity about xml:id as well.
... but I think we have to put a stake in the ground somewhere.

Paul: Doesn't XHTML have xml:id?

Alex: No, because they already have an attribute named 'id' that's of type ID
... SVG Tiny 1.2 has the same problem.
... and they had to describe a convoluted workaround for the browser vs. non-browser cases.

Norm: I wonder if we should remove xml:id from the recommended profile as well.

Alex: I'm not sure about that. I don't think the implementation is a huge issue. Documents which have both xml:id and id are, I think, pathological.
... If the performance degrades when you do that, don't do that.

Norm: Yeah, I'm not sure either, I'm just wondering.

Alex: I'd like it to stay because it's so useful for fragment identifiers.

Norm: Alex is going to look into XInclude in JavaScript which seems like it will have a bearing on what we think we've accomplished.
... There's been less resistance than I imagined to my fairly radical proposal. Maybe we should let it sit for a couple of weeks and revisit after we know the XInclude story.
... If we can get some clarity on xml:id in the browser in that period as well, that would be good.

Paul wonders if we could do better about attributes named "xml:id" and "id"


Summary of Action Items

[NEW] ACTION: Alex to investigate the possibility of doing XInclude in a conformant way only with JavaScript. [recorded in http://www.w3.org/2011/04/21-xproc-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2011/05/01 19:12:59 $