Publishing Working Group TPAC 2017 Summary

Part of Events

Author(s) and publish date


On November 6 and 7, the Publishing Working Group held a face-to-face meeting in Burlingame, California. The meeting was part of W3C’s massive TPAC conference, where immersion in spec work was the goal. Having TPAC at an airport hotel meant there were no distractions, which was perfect. We had participants from five continents, ranging in age from “don’t ask” all the way down to less than one.

Ivan Herman focuses his attention on a baby.

Ivan Herman, Garth Conboy, and a PWG guest

Update on FPWD for WP and PWP

The working group’s overarching goal is to publish first public working drafts (FPWD) of both the Web Publications (WP) spec, edited by Matt Garrish, and the Packaged Web Publications (PWP) spec, edited by David Wood. We hope to publish these in the next few weeks, and TPAC was the chance to discuss and hopefully resolve many of the major issues.

Of course, a specification without tests is just words, and we have lots of volunteers to help with testing, but no one (yet) to lead the testing effort.

Web Publication Lifecycle and Manifest

Naming things is hard—we even use the term “bikeshedding” to describe arguing about names. We’d previously used the term “manifest” to describe the components and structure of a web publication, but the term is so overused in other contexts that we decided to use “waybill” as a working term to avoid confusion for now.

One of those contexts is the Web Application Manifest (WAM). Through discussion in the group and with Kenneth Rohde Christiansen, we agreed that we will likely reuse some of the terms from WAM, and perhaps WAM will be able to adapt some terms from publishing. However, we have different processing models, and PWG may need to diverge from WAM when it comes to processing. The PWG will continue to work with WAM as we sort out the details of our metadata requirements. We hope we can contribute to their work, and they can contribute to our work.

Tim Cole suggested that we use terminology from FRBR to discuss the lifecycle of a WP: Discover, Identify, Select, Obtain.

  1. Discovery is finding the publication via search, reference or some other means.
  2. Identify roughly corresponds with the identifier and the addressability.
  3. Selection is a disambiguation process.
  4. Obtaining is the process of opening the publication.

We also talked about whether it would be possible for each component of a web publication to link to the “waybill”, or somehow indicate that it was part of a particular web publication. But this would prevent a component from being reused in other web publications, which is a major goal of some publishers. The discussion did help us clarify our understanding of the components of a publication.


Richard Ishida, Addison Phillips, and Fuqiao Xue of the Internationalization WG joined the group to discuss their work. The site includes information on how to participate, requirements, developer support, and education and outreach at The goal is to have all Working Groups and specs do self-review using the ongoing work and requirements. The I18n WG is looking for contributions to the Typography Matrix. Even photographs of books in the languages with gaps in the table are helpful.


The group has put together packaging requirements. EPUB is a packaging format for web content like HTML and CSS, but it doesn’t actually work directly in web browsers. What would a packaging format for web publications look like, when we want it to be a part of the Web? We discussed existing packaging formats on the Web, including MIME, webpack, tar, gzip, zip, and CBOR. Leonard Rosenthal gave a presentation about the future of PDF, which could potentially be used as a packaging method for web content in the future. We have not yet reached a conclusion about which packaging format we will specify, but the FPWD does not need to be that specific, and David Wood graciously agreed to edit the draft.

Synchronized Media

Marisa DeMeglio and Daniel Weck of the DAISY Consortium offered an overview of the work they have been doing to ensure that synchronized media publications continue beyond EPUB 3’s Media Overlays. Synchronized audio books usually offer sound with highlighted text as the user progresses through. Requirements and design options are available at The basic requirements are that audio playback is synchronized, navigate in the audio the same way one navigates in the text, escape complex structures (e.g. out of a table), some customization (e.g. don’t follow footnotes). There are several specifications today that enable some of the requirements. We are also hoping to close the gap between audiobooks and books that have an audio component. The group recognizes that this is a need for the Web as a whole, not just for publishing, and we are exploring work on existing standards, such as MPEG and TTML. We are planning to set up a Community Group devoted to this work. See also Romain Deltour’s slides about synchronized audio books.

Security, Privacy, and Integrity

The PWG plans to rely heavily on the basic security model of the Web. We rejected restrictive models such as AMP or even limiting JavaScript as was done in EPUB 3. This has proven to be an unsuccessful method. Some feel very strongly that privacy for publications must go beyond what is available today because users have different expectations from publications than of the Web. Others stressed that terms like “spying” are misleading because observing reading habits allows UAs to offer readers experiences such as keeping their place in a publication. The group discussed adding slots for signatures to ensure integrity of the content (that the publication has not changed since obtaining it). There is a need for user privacy as well. Readers have a much higher expectation of privacy within a reading environment than in a web app, and we plan to include a statement, perhaps more social than technical, in our document explaining that.


Tim Cole has been working on a Locators document that expands on the Web Annotations Selectors and States. He queried the group to determine whether we need to retain some of the unique attributes of EPUB CFI, the fragment identifier for EPUB. We discussed whether there should be a fragment id (no), whether we need side bias and text position selectors (including sortability), and selection of continuous and discontinuous embedded resources. We are seeking feedback from those systems that use CFI about whether their needs are met. We’re looking at you VitalSource.

Meeting with Accessible Platforms Architecture Group (written by Avneesh Singh)

The joint session of PWG Accessibility Task Force, Accessible Guidelines (Silver) Task Force and APA was addressed the incorporation of accessibility requirements specific to publishing to WCAG 3.0 (Silver) and way forward for Media Overlays specifications.

Topic 1: Incorporation of accessibility requirements specific to publishing in WCAG 3.0.

The Silver task force was happy to receive the publishing requirements compiled at

They were also briefed about the architectural issues in WCAG 2.0/2.1 due to which accessibility metadata was accepted as optional conformance in WCAG 2.1, while it is extremely important component of publications. The Silver task force stated that these issues will help them in creating a better design. APA conveyed their intent to propose the concept of pages to HTML working Group because it is a requirement for mobile devices also, and this would address the needs of publishing as well as of mobile users. Regarding timeline, it is a longer term work. Silver task force will be working on research and design in coming months, and they stated that actual work would pick up one year from now.

Topic 2: Exploring paths for moving ahead Media Overlays specifications.

The presentation was well accepted. APA group mentioned that they worked on somewhat similar requirements and developed a document some years ago at

The groups also discussed the existing technologies that may be useful i.e. TTML and Web VTT. The issue with these technologies is that the media is the master and text is associated to it. But in our case we need the text to be the master because the audio or video has to be an overlay. The structure and formatting should be provided by the publication, and media should be synchronized with it. So, TTML and WebVT are doing the reverse of that we need.

The advice was to go through second screen work, and explore other groups that may have the similar requirements. Some group members also suggested some ways in which TTML and Web VTT can be useful for our work. The broad view was to start a community group to explore the path ahead, and APA will help in communicating this work to other groups in W3C.

Breakout session on offlining

We had a packed room for a breakout session featuring Brady Duga, Benjamin Young, Jake Archibald, and Dave Cramer (via Skype) to discuss some basic questions what we need to offline publications and what the components of a publication are. We quickly concluded that ServiceWorkers should be able to do the job, but there is a great deal yet to be defined. Many of the questions that need to be answered have to do with the relationship of Web Publications to browsers. Jake made it clear to Publishers that browsers are really bad at guessing what to do with declarative markup and specifications that don’t explain behaviors. As a group, we are considering the need to create a JavaScipt Polyfill that provides a UI. This would not be provided by every publisher. Instead, it would be a universal, default UI.

Next steps

There are many open issues on GitHub. Please add your thoughts, and we will publish FPWD of WP and PWP within the coming weeks. Thank you to all who traveled to Burlingame and all who participated remotely.

Minutes and Drafts

Related RSS feed

Comments (0)

Comments for this post are closed.