<ivan> trackbot, start telcon
<scribe> scribenick: nickruffilo
<tzviya> https://www.w3.org/2016/11/28-dpub-minutes.html
Tzviya: "Any comments on the minutes from
last week?"
...: "Minutes approved. Before we move on, reminder: the DPUB ARIA needs
samples so that we can test the spec. If you have content that uses the
DPUB-ARIA terms or the terms that correspond, please pass that on to
Shane, Ivan, or myself. It would be very helpful, or else we cannot
proceed to the next part of TR for this document."
Leonard: "This is the new roles document
you're referring to?"
...: "Last week I had that document as a PDF-UA (accessibility for PDF)
and we hope to use it for PDF, so we hope it will give additional
support for that and implementation."
Tzviya: "If you have samples, please pass that along to us."
Tzviya: "Moving on to the main agenda items - locators and web publications. i put 4 links in the agenda. We've been dancing around this topic for a while."
Ivan: "Only to say that what is currently in
the document - I've re-organized the document. What is there came out of
an earlier task-force discussion. That was before the issue-storm of
last september. That discussion, for those who came in since, has
changed a little bit. And so that part needs a very thorough re-write.
What is in the document now - there is only one change earlier. In the
locator discussion, we had quite a long part on - the question is where
do we find and how do we create the manifest, without defining what the
manifest is. And how do we do it. We came to a relatively complicated
approach to allow the manifest to be merged. We had some good reasons
for that, but the discussions with the web manifest people, and
ourselves, we realized that it sort-of works but it's complicated, so we
abandoned that."
...: "So i removed that part, and the only thing there is the possible
to have the manifest provide the link. I think that the whole section
has to be reconsidered. Thats all I can say right now."
Leonard: "For now, we should probably remove that section entirely from the document. When we think about the technical details of the manifest, these things will be more relevant when we know what the manifest will look like."
<tzviya> http://w3c.github.io/dpub-pwp/#locator-pwp-func
Tzviya: "Are you talking about canonical
locators of the PWP and not the UCR document."
...: "So we're talking about removing the 2.2.2 section about canonical
locators."
Ivan: "I'm torn on this - there are some issues and discussions about issues that are to be considered - and yes there ahve been discussions about what formats the manifest will look like. The fact of having the issues and the discussion somehow documented so that the working group has something to start with, I think that will be useful. If we could at least summarize what you, hadrien, I, have had in the past few weeks as to the problems, and the roles of the canonical locator, it's useful."
Leonard: "Part of the problem is that we merged 'state-locators' and state-independent-locators' together. The problematic ones are the state-independent locators - possibly move to 'huge problem we can't solve yet.' The state-locators are important to incorporate here."
Nick: "Can a locators solution be manifest independent?"
<leonardr> @nick - excellent question, and it goes (IMO) towards the state vs. state-independent difference.
Dave: "I've found a fair amount of this discussion over the years now to be somewhat frustrating. I find this very hard to talk about in the abstract. We might not even really have some fundamental agreement on what these web publications are. There was an interesting email from Ivan - contrasting his view as a web publication as this very roughly 'exploded epub on the server' VS what might be - possibly hadrien's view - a collection of web resources existing on different origins. Getting at some of those fundamental issues would be helpful. There was some good discussion about URLs about publisher P and library L and user M emails something to another user - we were making progress during that discussion."
<dauwhe> https://gist.github.com/dauwhe/4ba2c1a44c624a9d7ffb1b54aaf5d8be
<laudrain> +1
...: "As i was being frustrated, I wrote up a bit of a rant, that I put
into gist (link about). I'm sure there's alot here that's attackable but
I wanted to make the point that a publication is more than a collection
of links - the table of contents is not the book. Emphasis the role and
act of publishing - as we're going to have to structure our security
model around that."
Ivan: "Two things - the reason I put myself
on the queue is that I've already removed from the document the whole
story of states. This was one outcome of the discussion. You are right,
everything related to this state issue has to go. Also reacting to what
dave said - yes, because we had all those discussions on GitHub, I don't
want to lose that."
... "My point is very practical - the main role for this document, is
that it is an input for a much more serious standardization work. I
don't want to lose anything that will be on that working group later to
move ahead. "
Leonard: "Ivan, I think you're right. I'm OK - there is a way we can tweak what is there to make it less controversial. I've been out for a while, and I just re-read it quickly, let me take a look and give it deep thought. We can possibly come up with a version that we can all agree on."
Hadrien: "I wanted to say that I actually agree that locators are not necessarily related to the manifest itself. I don't think it's related even, with the state of the publication. Packaged, unpackaged, on the web, it doesn't matter that much, or how the manifest is serialized. We should have broad agreement on how we define & reference various items and that would be useful, and that's doable with the current work being done. This question will also come on the table very soon on the readium level. If we can get to a point where both organizations agree on the principles, that would be a good thing.
Tzviya: "One of the things that has been complicated about this discussion is that terminology we're using has been confusing. And it's plagued us along the way. We use the work 'canonical' loosely - such as a canonical URL or in a looser way. We talk about a PWP-Processor, which isn't something we should talk about. Maybe call it user-agent? It's some terminology that makes this hard to swallow. Please look at the Gist that Dave put up. It's some basic principles. We should have a discussion on principles and not just assume we agree. And maybe we want to go back to basics."
Leonard: "Fully support going back to principals. The locator group chose PWP-processor over User Agent was that we didn't expect the User Agent that was going to manage to locators. IN the original locators document, we came up with a list of use-cases where the processor wasn't necessarily a UA. We really did, purposefully, think that there is a separate thing from a UA that will be handling the locators."
Ivan: "Let me add to this - the document elsewhere still carries the term 'web publication processor' simply because we don't know at this moment will be right-away implemented by browsers. I do not think it will at first. I expect whatever we do will require an extension or add-on. We need somehow a level of abstraction to talk about this. For that, I believe that having a term of a separate processor is useful."
ivan: " maybe an extension - we have to have a processing entity that is not necessarily part of the User Agent as we refer to it. Mainly the browser."
Tvziya: "Ivan you probably know this better than we do, but we should see how things are used across W3C. I think User-Agent is used broadly, even when things like assistive technology comes in."
Ivan: "I agree. we should check that"
<tzviya> acl le
Nick: "We should write specs for the end-goal."
Leonard: "We want user-agent for the wide definition of User-agent to fully understand PWP - packaged or not. So we don't talk about browsers, we talk about user-agents as the general case."
Tzviya: "Any other thoughts on locators and mapping?"
Ivan: "I would like to see actions defined -
that go to people other than Ivan."
... "I see 3 actions - not of equal weight. One is 'I yield to the
user-agent story, so the doc has to be cleaned up to talk only 1
process. I volunteer for that. There is another action - to take Dave's
comments and add them as part of the document. One of those things I
don't want lost. I don't volunteer for that. The locator section should
be re-written documenting the discussions we've had on github and I
prefer that not to be me. "
<leonardr> I am willing to take on the Locator (or other) sections
<tzviya> https://gist.github.com/dauwhe/4ba2c1a44c624a9d7ffb1b54aaf5d8be
<leonardr> and i need to drop - sorry - talk to everyone online! (and just put my assignments in the minutes). Thanks!
Tzviya: "Please review again the rant. No one commented on Locators. "
<ivan> NickRuffilo +1
Deborah: "A couple of things - I think this is all good discussion. One thing we're cycling through here, and heard constantly - is that a publication can be defined as a collection of links. So we need to make up some words and stick with it. We have to define things - that will not be inclusive of everything. It is OK that your definitions hold things."
DavE: "I will offer to add my rant into the PWP."
tzivya: "How do we make this human readable"
Ivan: "For now, lets concentrate on locators - lets take one thing at a time."
Tzviya: "Like ivan said - lets stick to this section. We want human readable - but we don't want to lose meat. It's still technical spec."
Nick: "I'll give a go before next week."
Ivan: "For me to go through and change the processor and these sort of things - affect the whole document, and things get more complicated, and I'll try to do that tomorrow so it's out of the way and doesn't interefere with what you're doing."
Tzviya: "Dave, by next week?"
Dave: "Yes."
Tzviya: "Any other thoughts on locators?"
all: :: crickets ::
Tzviya: "I forgot to ask availability. Next week we assume people are relatively available. 19th?"
<dkaplan3> =1
<clapierre1> dec 12, 19 +1
<ivan> +1
<laudrain> Not available for the 3 next Monday…
<dauwhe> ~1
<clapierre1> -1 26
Tzviya: "26th - cancel?"
<dauwhe> -infinity for December 26th
Tzviya: "Anything else?"
davE: "Packaging? There was a resurrecting - and now Google is talking about extending it. And they talk about packaging signing."
<tzviya> https://discourse.wicg.io/t/proposal-packaging-for-the-web-signed-and-indexed/1827
<dauwhe> https://github.com/dimich-g/webpackage/blob/master/README.md
Tzviya: "Link to explainer above."
Dave: "I've been working on implementing this in my sample reader. Things can now be downloaded into a package format. The types of serialization details are a fundamental question. "
Ivan: "In the meantime - on the one hand it's interesting and good to know it's happening. I wonder if it affects us right now. In a sense, our main concern is to understand and handle the web publication and a packaged web publication is important, we know that, but how the packaging is done - the kind of secondary issue, depending on market forces, etc. Is there something very fundamental in the web packaging stuff that makes it very different than what epub is using today?"
DavE: "I think there is - maintaining the idea of the origin and signing the package with the origin certificate. "
Ivan: "I don't think anyone has thought of it in terms of epub?"
Dave: "I don't think so. And my guess is
that it isn't different than epub much."
... "I see it as a way to address one of the fundamental issues/items we
have."
Ivan: "I understand that, but playing devils advocate, my understanding is that the current epub packaging is signed, etc, it can be done. It's not specific to this packaging format."
Dave: "That's right, but I don't think epub has had that idea of tying it into the web security model."
Tzviya: "When you ahve a major browser researching a packaging method - it's worth going along for the ride."
Ivan: "We may go along saying, in a working group, that the package doesn't specify which packaging format or syntax is used. One profile may use one or the other - i'm making it up as we go..."
Dave: "I want us to remain concious to this idea to ensure what we define is compatible with this."
Tzviya: "Anything else?"
<ivan> https://discourse.wicg.io/c/dpub
Ivan: "We discussed a few weeks ago - we set up discourse entry for digital publishing, i just need to find a URL. there is now a category that has been set up so that if we want to discuss things that are digital publishing but also useful to others in the web browser world, it's a good place to have technical discussions. It's good to know it's possible."
Dave: "My only concern about that is that - as far as I know, a discussion there cannot have more than one category - and that causes concern. I found no way to assign multiple categories, and is a huge problem for us. "
Tzviya: "Who runs that? Robin started it but who runs it?"
Dave: "Maybe we should start a discussion about having multiple categories :)"
<darobin> the WICG runs it (ie its chairs, normally with a light hand)
<astearns> chairs are Marcos Caceres, Chris Wilson, and Yoav Weiss
<darobin> WICG -> https://www.w3.org/community/wicg/
<darobin> see also current specs that it's working on -> https://wicg.io/