Publishing Working Group Telco — Minutes

Date: 2018-11-19

See also the Agenda and the IRC Log

Attendees

Present: Tzviya Siegman, Wolfgang Schindler, Jeff Buehler, Leonard Rosenthol, Wendy Reid, Rachel Comerford, Ivan Herman, Nick Ruffilo, George Kerscher, Joshua Pyle, Luc Audrain, Jun Gamou, Zheng Xu (Jeff), Avneesh Singh, Marisa DeMeglio, Dave Cramer, Franco Alvarado, Romain Deltour, Matt Garrish, Chris Maden, Benjamin Young, Ben Schroeter, Gregorio Pellegrino, Mustapha Lazrek, Brady Duga, Daniel Weck, Charles LaPierre

Regrets: Ric Wright, Vladimir Levantovsky, Tim Cole, Caroline Hayes, Garth Conboy, Teenya Franklin

Guests:

Chair: Tzviya Siegman

Scribe(s): Nick Ruffilo

Content:

Tzviya Siegman: Last week’s minutes - comments?

Tzviya Siegman: https://www.w3.org/publishing/groups/publ-wg/Meetings/Minutes/2018/2018-11-12-pwg.html

Resolution #1: last week’s minutes accepted

1. F2F meeting

Tzviya Siegman: Last week - after the meeting - we mentioned that we’re planning a spring face-to-face in Cambridge Mass at Google’s office. This is our tentative date - 6th and 7th of May. It’s graduation season, so hotels can be pricey.
… We’re working with the google co-ordinator, but Air BNB might work for people.

Marisa DeMeglio: Is there an address so we can look?

Tzviya Siegman: Not far from Kendall square - but very close to public transportation. Next call we’ll have more details.

Ivan Herman: Map for the kendall sq. google: https://goo.gl/maps/MQsWTNtDA792

2. Agenda format suggestion

Wendy Reid: We’ve proposed a 2-week in advanced agenda format, so the chairs will propose 2 agendas (sent on thursdays) in conjunction with earlier agenda, we will also provide explainers on the agenda, so whatever topic we plan on discussing will have context.

Tzviya Siegman: If anyone wants to help do the summaries in GitHub - please let us know. It is quite a bit of work and we need help.

3. Use cases

Tzviya Siegman: First item is use cases and the MVP. Franco - can you talk?

Tzviya Siegman: UCR: https://w3c.github.io/dpub-pwp-ucr/

Franco Alvarado: I’ve completed the mapping to the WP spec for requirements. Each requirements have an OL with references to the WP spec. For the use case requirements I’ve add an additional requirement - around the TOC, if anyone has things they’d like to add about that.
… We were talking about the structure of the document itself - adding IDs, data-attributes - we can sort them and gather them into a table to indicate the types of cases and which area of the publishing industry or items strictly to the User Agent.
… What leads me to the MVP is - Josh had the idea of tier-ing, but we need to get to the MVP first. Related to MVP - how - are we talking about the user agents or the WP itself? Is it how we want the UA to react with the WP, or how we want to talk about it…
… Where would the affordances link to here? etc..

Joshua Pyle: I actually got these in just seconds before the meeting but based on conversations I’ve had, I’ve gotten some feedback from Ivan and others, I’ve abandoned the numbered tiers… (if you look at this link…)

Joshua Pyle: https://cdn.staticaly.com/gh/w3c/dpub-pwp-ucr/conformance-tiers/index.html#conformance-tiers

Joshua Pyle: This link shows the new, easier to understand, proposal. Have minimal requirements that correspond to the MPV then 3 groups of others. They aren’t necessarily stackable tiers, you could have minimal + any of the optionals. They don’t necessarily have to be stacked, etc.
… You could just support minimal + packaging, or minimal + customization, etc. That allows us better ways of thinking about things. One question Franco just asked was - does this apply to the document or the user agent…
… My answer is that they apply to the user agent and how it conforms to the WP. You can have a WP that includes tons of functionality, but the conformance is how much the UA supports.

Ivan Herman: +1 to josh

Zheng Xu (Jeff): I agree with the MVP. We need to keep aware that the features of MVP are testable through a UA. Shrinking an MVP to make it a baby step - if we can define a way that UA can switch to reading mode…
… Even though we can define progress, but to make it take the w3c approach and recommendations to RC, this will make it more successful, and we can recieve feedback from UA or publishers, then we can make the next step.
… A bigger MVP or extended. I recommend we start small. The other thing is that I don’t feel we need to make something close to epub. A publisher can read the specification and make what they need. I don’t think we can catch up with epub that soon.
… A baby-step can let the UA switch to reading mode and be testable for each UA.

Tzivya: Maybe we should have done MVP then use cases.

Benjamin Young: https://www.w3.org/TR/html/infrastructure.html#conforming-document

Benjamin Young: Josh - I like the new label levels than the numbers. HTML 5 (just linked) has conformance classes that are even more described - I’m wondering how much of this could help with the decisions were making now. It’s helpful when thinking about what might affect certain clients and not other clients.
… These put things into document - or dismiss them. That might be helpful.

Joshua Pyle: +1 to HTML conformance example

Tzviya Siegman: Before switching to MVP - You’ve done great work, but it sounds like it’s separate. It might be helpful to just get on the phone and ensure you’re working towards the same goal. Sounds like you’re working on other parts, and not conflicting, but worth going in the same direction.

4. Minimal Viable Product (MVP)

Tzviya Siegman: On the MVP - I went through the TPAC meeting minutes. I pulled together the notes on what people said they felt was necessary for the minimum. One thing we need to define - we’ve had questions about where we’re opening this file…
… Many people - especially George, about the 1) Omnipresent TOC always accessible. 2) ability to navigate through default reading order, 3) being able to nagivate in the right direction (left-right, vertical, etc)
… Not brought up in TPAC and possibly not part of TPAC is the addressability (a URL for the whole publication or sub-resources)

Ivan Herman: I was looking at the Minimal notes that Josh did and it nicely maps to what we did at TPAC besides one thing. The TOC example was done by Franco separately, so it hasn’t been merged it. It should appear as part of the minimal.
… The minimal thing is pretty much inline with what we’ve said.

Tzviya Siegman: We might need to go back and review… Perhaps this is a good way to look at what we need. We’ve been trying to figure out what is lacking - such as ‘web publications should have all functionality of the web’ and that they should have all metadata elements.

Tzviya Siegman: https://cdn.staticaly.com/gh/w3c/dpub-pwp-ucr/conformance-tiers/index.html#conformance-tiers

Tzviya Siegman: Discoverability is something we have not addressed - it’s an interesting discussion…

Ivan Herman: I don’t understand what must - if the web publication must be easily discovered, it has a URL and we have that, but differentiating between essential and non-essential, i’m not sure where we noted that in the UCR

Nick Ruffilo: I think the only think that makes sense as a requirement for discoverability is a URL
… I don’t think we need to define SEO

Dave Cramer: http://www.sagehill.net/docbookxsl/

Dave Cramer: I’ve talked to some web people who’ve looked at our requirements pages and I fear they will look at our MVP and say ‘the web can already do all of this’ (see link).
… It meets all requirements - every page has a nice button at the bottom that takes you to the TOC. The arrow graphics to get to next and previous. I think we need a better story than this to convince people that we need new features on the web platform

Joshua Pyle: Maybe it’s not required for MVP, but one more compelling use case in requirement 4 is having to do with offline-ability. Some young lady wants stock prices, but when she’s offline, the stock price doesn’t need to be read, but she can still read the publication. That’s kind of an ambiguous rule. It should be said, I haven’t really criticized the use cases, just re-organized.
… That specific requirement might be ambiguous or unnecessary.

Ivan Herman: https://hpbn.co/

Dave Cramer: https://resilientwebdesign.com

Ivan Herman: To try to answer Dave - yes, there are lots of things that can be done; see also, e.g., another example. I usually put a different example (link pasted). It can be offlinable. You can do publications on the web today that can do all of these things. With high-performance browsing, it’s a good example. But, in order to do that today, you have to do relatively complex work (the publisher has to) that is highly specific to the publication.
… There is javascript, service workers, and all that jazz. It lists all the HTML documents, etc. To make it this way, it just doesn’t scale. What we’re trying to do is to make this process as standard as possible, it would be much easier and standard to create books like this.
… Yes it’s doable but it requires three PHDs and Masters to do it…

Dave Cramer: I agree, but they’ll come back with ‘every website has to do this’ and ‘you can just make a framework’ I’m worried that our message isn’t going to be compelling to the browser vendors.

Tzviya Siegman: we need to think about things like offlining - if it’s something we want to be an MVP. Whichever tool needs to determine what items are and are not essential when it comes to offlining as well.

Benjamin Young: The core component of what Dave is saying is that there are web-publications already on the web. They all share similar needs and have similar desires, like an immersive reading experience where you stay within the bounds of the publication.
… There are commonality among all of them - along with the simple things we’d want as humans to read things on the web. What do people publishing to the web do now, and what can we offer them to make it simpler.
… There are lots of things we can do. How we get them into the web platform, that’s the question. Readers want to be immersed. They (authors/publishers) want to package it and not have to worry about it once it has been delivered. Those are the pain points.
… Can they be done - yes. Can we make it easier/reduce cost, yes…

Tzviya Siegman: Taking a look at the candidates for requirements in the MVP. But, the use case about offlining (requirement 4) it should be removed.

Joshua Pyle: I’ll clean it up and remove it.

Tzviya Siegman: 7 - data should be considered a resource as text and images is.
… 8 - we need to uniquely define a web publication (addressability)
… Req 9: UA must treat a WP as a single item.

Benjamin Young: With 8, a unique identifier, are we talking about a unique identifier for the web pub, or something that uniquely defines it as a web publication?
… Yes, we have a URL, but does it come back as something unique that lets you know it’s a web publication?

Ivan Herman: +1 tzviya

Tzviya Siegman: I think if we word it as uniquely addressed as a web publication - that clarifies a bit. So it’ll be changed from identify to addressed
… 9 - this would be when we start expanding UAs.
… 10 - all constituents and their resources need to be addressable. Tim did significant work on this years ago and may want to go back to it. We do not have a solution now for this.

Tzviya Siegman: If items are unpackaged, it’s not an issue. If it is a packaged situation or superstructure, then it’s an issue. Same with an epub, you needed to create a CFI

Nick Ruffilo: is packaging part of MVP?

Joshua Pyle: If i’m publishing a web page, the images need addressable URLs. I don’t see the confusion here. Everything that links to it - or it links to - needs it’s own address. It might be so simple we don’t need to say it.

Benjamin Young: I think that uniquely identifying is handled by URLs. What is not handled is pointing at “this image within the bound web publication” but to say I’m pointing at the chapter 3 resource within the book, that you become “in the book” and not just the resource. The annotation stuff Tim was exploring was part of that. Epub’s packaging prevents this as none of the resources are addressable on their own.
… other packaging options may prevent that as well. If I link you to the logo of a web-app, that wouldn’t be known to be part of a web app.

Dave Cramer: A lot of these requirements are both obvious and impossible. It’s obvious that an image is going to have a URL, but the impossible part is to say that the image is part of some more complex structure.
… or an actual user requirement would be needed. It’s a referencing requirement - but I don’t think it’s the final requirement, it’s something we think is a solution of the requirement.

Laurent Le Meur: the text of the requirement is about stability. If you look at the 2 usage examples, it’s about how to reference something in a stable way. It’s just paraphrasing the good URI on the web. It’s about good practices, not about requirements.

Tzviya Siegman: This too needs to be re-written but should be moved out of the requirements.

Benjamin Young: One thing that haunts me as a publisher is 404. What the URLs concept is fine, it’s a rental economy. It will change… If we change anything it would be the stability. I could put something out and you could depend on it forever. That’s part identification, but that’s the kind of thing I’d like to see us address.

Tzviya Siegman: 11: for purpose of layout, each resource is it’s own document. Some people say they want this, some say they don’t want it. Some people say they just want CSS for some items, some say they want all items. We have to figure out what we want here. For counters, we may want it applied to everything. For this requirement, we want each resource treated as a separate document…
… we have to make rules for everyone, or we need to create a meta-rule that allows it to apply ‘globally’

Joshua Pyle: The term resource here is very confusing. An image (figure) in one chapter of a book should not be treated as a separate content. The intent is that chapter 1 and chapter 2 should not be mixed, so maybe we need a more specific term for a resource that means a document or sub-document. The term resource is way too inclusive for this restriction

Ivan Herman: I don’t really understand this requirement. With what you said with the CSS - if I author a book with several chapters, it’s up to me to which CSS I refer to, and the global one will be the global one - but that duality already exists. There is a question of the counter and I’m not sure how that should be address, and it’s likely not a minimal viable thing, but what is here, in this requirement…
… as I understand it, can be handled by this today…

Tzviya Siegman: I’m not sure how this made it in…

Brady Duga: Regardless of what people want - and I understand counters across documents would be great, but it’ll never happen

Zheng Xu (Jeff): Almost the same. If publishers don’t create content in a specific way, lets not mix the concepts. CSS group would just ask “what is content created this way?” The reading system shouldn’t really have to do to handle this.

Dave Cramer: I agree. There is historical items here - where epub has put a bunch of documents in a single window so you could scroll through all content, but it’s problematic when it comes to the web, which is why we’ve had statements like this.

Tzviya Siegman: Taking away the history - I think that, if that’s the reason, that falls into the category of ‘personalization.’ If we wanted to include that, it should be re-written.

Joshua Pyle: Based on everyone’s comments, I proposed that rec 11 is re-written, and the only use case is about counters. I propose we kill it.

Ivan Herman: It would be milder - we shouldn’t remove it altogether - but make it clearer that we’re referring to counters and continuity of them, but it would go into extended (or very extended) It would require some very special css.

Joshua Pyle: I’ll move 11 to extended, but someone else who is more familiar with the intent of this requirement please edit it?

Tzviya Siegman: we have just a few minutes. 12 - there should be a means to navigate the preferred reading structure. Look at the TOC requirements and see if you can make sense of this. Look at navigating from the default reading order.
… We talked about that as an MVP, but any other item is outside of MVP. So if it’s right-to-left, the whole book is right-to-left. If you have any other requirements, be in touch with Franco or Josh or open an issue in the GitHub.

Ivan Herman: Once this is done, we should spend some time on the most boring thing - to reconcile the terminology between WP and this document. ‘constituent resources’ yet we might be using other terms - we need to figure out our terminology. Lets have the whole structure done, then republish the note, etc.

Tzviya Siegman: We had proposed creating an MVP task force. We got pretty far today. Josh and Franco - would you like people to help with the UC document?

Zheng Xu (Jeff): I can help if needed for MVP

5. Resolutions