Publishing Working Group Telco — Minutes

Date: 2017-10-16

See also the Agenda and the IRC Log

Attendees

Present: Ivan Herman, Dave Cramer, George Kerscher, Wolfgang Schindler, Avneesh Singh, Matt Garrish, Tzviya Siegman, Tim Cole, JunGamo, Chris Maden, Lillian Sullam, Baldur Bjarnason, Ric Wright, Rachel Comerford, Joshua Pyle, Luc Audrain, Romain Deltour, Mateus Teixeira, Laurent Le Meur, Vladimir Levantovsky, Jeff Printy, Bill Kasdorf, Ben Schroeter, Brady Duga, Garth Conboy, Hadrien Gardeur, Benjamin Young, Gregorio Pellegrino, Leslie Hulse, Reinaldo Ferraz, Ben Dugas

Regrets: Marisa DeMeglio, Charles LaPierre, Deborah Kaplan, Peter Krautzberger, Leonard Rosenthol

Guests: Marty Picco

Chair: Garth Conboy

Scribe(s): Dave Cramer

Content:

Tzviya Siegman: let’s get started

Dave Cramer: [ping pong chime]

Tzviya Siegman: https://www.w3.org/publishing/groups/publ-wg/Meetings/Minutes/2017/2017-10-02-minutes.html

1. last minutes

Tzviya Siegman: any comments on the minutes?

Resolution #1: minutes approved

Tzviya Siegman: minutes approved

Tzviya Siegman: any new members this week?

2. new members

Dave Cramer: MarthP this is Marty from Atypon, which is part of Wiley

3. spring f2f dates

Tzviya Siegman: we have a form for possible dates for the spring F2F

Tzviya Siegman: https://docs.google.com/forms/d/e/1FAIpQLSc0YlroocKuvfP0bd5pNRFUChY7QJ2gYGOkE7zx-tSSpkBxkg/viewform?c=0&w=1&usp=mail_form_link

Tzviya Siegman: please fill out the form

Garth Conboy: there’s no time like the present, so if you haven’t filled out the form please do so now!
… we will probably discuss this on Wednesday, and have a final decision soon

4. Lifecylcle of a WP

Matt Garrish: the lifecycle is picking up from where we left off two weeks ago
… we don’t want to proscribe a solution for the FPWD
… I’m just taking the issues that have surfaced all along

Matt Garrish: so I’m trying to lay out the general topics in an editor’s note
… so we don’t have a big gap in our spec
… the first part is basic principles… we’re not trying to break or fork the web
… and that a web publication shouldn’t have to be it’s own app
… the second part is a breakdown of what used to be in this section
… the issues and problems around initializing the WP

Ivan Herman: The version of Matt’s contribution in the PR: https://s3.amazonaws.com/pr-preview/w3c/wpub/lifecycle.html

Matt Garrish: taking it offline, searching across documents, how do you traverse history…
… updating a manifest… these are issues we know we will face
… this is what’s on our plate
… do we integrate this? Should we list more? How should we get feedback? Broad issues in GitHub haven’t worked well in the past.
… we want a wide range of people to voice their opinions
… we should put something in SOTD about what we want feedback on
… fin

Garth Conboy: any feedback right now for matt?
… consensus to accept the PR?

Ivan Herman: first, I think the PR should be merged
… as for getting more feedback, TPAC isn’t far away, so we will discuss there
… and then make a lot of noise when the FPWD is out, maybe right after TPAC

Tzviya Siegman: I’m in the ship-it category
… once we publish, we might want to contact specific people, and ask for feedback
… like the browserfolk

Garth Conboy: that sounds good to me

Luc Audrain: +1

Garth Conboy: Matt, Make It So (in Picard voice)

Resolution #2: Merge #84

Garth Conboy: Leonard is next?

Ivan Herman: He had an emergency

Garth Conboy: we can move this to next week

Garth Conboy: next is timCole on selectors

5. Locators

Tzviya Siegman: https://w3c.github.io/publ-loc/

Tim Cole: https://github.com/w3c/wpub/pull/75

Tim Cole: there’s a PR associated with this discussion about internal locators
… how to identify a fragment or location in a wp
… we tried to put that in WPUB doc, but it was too long
… so we started a new document
… so Ivan wants to introduce the context

Ivan Herman: for those who didn’t read this or its predecessor
… this is the selector model from the web anno WG
… we should be able to point to an area in a resource that might not have an ID
… bookmarking or annotation is the obvious example
… that goal is the same as for epub cfi
… but the group took a different approach
… because a single solution wouldn’t work
… so the group developed a JSON object that included various ways of IDing a point
… could be ID, could be text match, could be character index, could be range
… the idea was that IDing these things as JSON, an annotator can handle that
… we were looking at it for WP
… we think it could solve the use cases
… 2 issues: we had to expand the web anno model

Tim Cole: https://github.com/w3c/publ-loc/issues

Ivan Herman: if we have a WP we might need a range spanning multiple docs
… the other issue is more general
… EPUB CFI was a fragment identifier
… web anno is a json structure and not a fragment identifier
… there’s a way to encode the JSON into a fragment identifier
… the Q is, do we really need that?
… those fragment identifiers are even harder to read than CFI
… and we also have to go through an IANA registration… there are lots of admin hoops
… for the use cases for WP, do we really need fragment identifiers
… or can we use the JSON structure, knowing that annotation systems already have adopted these
… there are also some issues around details

Tim Cole: I want to go thru a couple of use cases in CFI

Romain Deltour: to clarify, one of the reqs for CFI was that they could be sorted
… you could tell if one was after the other, you could sort in doc order
… that isn’t possible with this proposal
… but I don’t think it’s important for our needs

Ivan Herman: I agree with both—it can’t be done, and it’s not terribly important

Tim Cole: the Q is how fixed is the reading order?

Tzviya Siegman: why does it always come back to that? :)

Garth Conboy: we do have dependency on the reading order for ranges crossing docs

Tim Cole: we suggest you just list the resources and their segments
… it assumes that there’s some reason they are grouped

Rachel Comerford: the issue across docs carries into the higher ed market, this is a problem we have with CFI
… we can’t make CFIs that cross HTML files

Garth Conboy: I remember that ranges were done with two pointers, which implies knowing the reading order

Tim Cole: in the original webanno model, there was a range indicator for within a single file
… we’re looking now at multi-resource selector where you get identification thru any of the other selectors
… I don’t know how well this would work, because we haven’t tried

Garth Conboy: so if you’re starting in doc 3 and ending in doc ten, you’d have to reference (4..9)

Tim Cole: yep
… that’s new

Tim Cole: don’t know if it will be good enough for Rachel’s use case. We should talk at TPAC

Garth Conboy: this is potentially problematic with multiple reading orders

Tim Cole: https://github.com/w3c/publ-loc/issues

Tim Cole: if you go to the link, and look at the issues
… the frag ID question Ivan already mentioned
… in the web anno model, we’re always talking about a chunk of content
… everything is a segment of a file
… in CFI, there’s location and there is side bias
… do you need to specify a position between char 6 and 7, but bias to the start of 7 rather than the end of 6?
… is that a real use case?

Garth Conboy: I don’t remember

Tzviya Siegman: maybe SVG?

Tim Cole: page breaks?

Matt Garrish: if the char boundary is a page boundary, you need to know where to stay

Tim Cole: we havent figured out the side bias yet

Tzviya Siegman: could you work around that with anno selectors

Ivan Herman: it’s a matter of really understanding the use case

Tim Cole: might require an additional property
… my hope is that the anno model will fit

Garth Conboy: I do recall pagination as being one of the reasons

Tzviya Siegman: I had 2 comments
… one was about range, indexing is a use case
… people need page numbers, but even that has ranges, and may cross documents
… the other comment is on fragment identifiers, would this be odd without a new media type?

Ivan Herman: odd is not a good term
… formally speaking, it’s not possible
… frag IDs are defined for A media type
… if we want to have frag IDs, then we have lots of administrative problems
… and I don’t want to go through that again :)
… the range story… the way it works in one doc, is defining the start and the end of the range, using other selectors
… its a continuous set
… you may start in paragraph 1 and end in para 21, and it’s the whole range
… can you define those ranges over doc boundaries? We still have to discuss that
… I want to emphasize that we’ve tried so far to be as close to the web anno model as possible
… we shouldn’t redefine anything
… we might have to add stuff
… there is already a REC and implementations
… and we don’t want to break existing implementations

Tim Cole: regarding fragment id
… there is precedent in media fragments
… which works for multiple mime types
… CFI is tied to the manifest of an EPUB
… you can be trying to talk about a segment of a resource that’s inside another resource
… like a segment of an image that is in an HTML page
… but there is a scope, the embedding html page
… so with web pubs we have an html file that’s part of a WP
… so again we have a scope
… so we might need this idea
… the last idea is kindof related
… if I have a resource that appears in multiple places in a publication
… do I need to specifiy which instance I’m talking about? the same image in ch5 vs ch9?

Garth Conboy: we didn’t have that problem in CFI because we didn’t allow it to happen
… CFI postdated that requirement
… the other Q
… the proposal now works for WP
… would this work for PWP?

Tim Cole: I haven’t thought about it
… in WP, we think everything should have URL
… if in PWP, the identifiers are only available via the manifest, then we might have an issue

Tim Cole: we should use canonical identifiers
… maybe it’s a must

Ivan Herman: there’s a concept of embedded resource selector
… by default each resource has it’s own URL
… in some cases you want to say that this URL has to be understood in the context of a WP
… so we have 2 diff ways of expressing that
… if you use that, if it is the package a package has a URN
… each constituent also has an identifier
… this mechanism should be usable to solve this issue
… one more thing
… we would need to have an agreement that we merge the PR

Tim Cole: https://github.com/w3c/wpub/pull/75

Ivan Herman: so that this is now integral part of the work
… even if it’s in a different doc

Bill Kasdorf: we do have to keep a11y issues in mind
… if you have a fragment of an image, how do you manage the alt text so it’s still accessible

Tim Cole: the fragment wouldn’t include the description

Ivan Herman: I want to see the use case
… this is to put annotations or create bookmarks, that’s what it’s all about
… I don’t think there’s an a11y scenario here

Bill Kasdorf: I think you’re right

Ivan Herman: this did not come up in web anno for these reasons

Brady Duga: it seems like the use here would be a picture, I draw a circle around someone’s face, and say something
… you’d want alt text to explain what’s circled
… if it’s a photo of the class of 1987

Ivan Herman: the alt text is in the document, part of the HTML

Benjamin Young: you’d need to annotate the alt text—not the photo

Brady Duga: the alt text describes the entirety of the picture, not the fragment

Ivan Herman: you make an anno on the picture, and then you add the alt to the anno

Tzviya Siegman: I don’t want to go too far down this rabbit hole
… brady, you’re talking about the a11y of annotations
… and we have two more items on the agenda

Garth Conboy: Tim, were there other issues?

Tim Cole: no

Garth Conboy: going back to Ivan’s suggestion of merging this PR
… seems like a good plan to me

Tzviya Siegman: +1

Resolution #3: merge PR #75

Garth Conboy: let’s do that

Tim Cole: I’m hoping to have this doc to discuss at TPAC

6. Pagination issues

Garth Conboy: https://github.com/w3c/wpub/pull/65

Tzviya Siegman: https://github.com/w3c/wpub/pull/85

Proposed resolution: close #65 and merge #85 (Ivan Herman)

Ivan Herman: I saw Florian’s comments
… he was commenting on Dave’s changes on his PR
… I would be in favor of merging things
… the comments of florian are valid and should be discussed
… but it’s good to have a complete FPWD

Garth Conboy: I would agree with Ivan’s proposal
… any objections?

Resolution #4: close #65 and merge #85

7. TPAC Agenda

Garth Conboy: there are references to a fleshed-out agenda
… now everyone can comment but not edit
… I encourage folks to review and comment

Tzviya Siegman: We have more requests at joint meetings

Garth Conboy: anything else to bring up?

Dave Cramer: (silence)

Garth Conboy: https://docs.google.com/forms/d/e/1FAIpQLSc0YlroocKuvfP0bd5pNRFUChY7QJ2gYGOkE7zx-tSSpkBxkg/viewform?c=0&w=1&usp=mail_form_link

Garth Conboy: everyone, take the last five minutes to fill out the form for the Spring F2F

Wolfgang Schindler: have a nice afternoon!

Garth Conboy: that’s all, folks!

8. Resolutions