Publishing WG Telco — Minutes

Date: 2017-07-10

See also the Agenda and the IRC Log


Present: Luc Audrain, Tim Cole, Vladimir Levantovsky, Mateus Teixeira, Dave Cramer, Rachel Comerford, Tzviya Siegman, Ric Wright, Harriett Green, Garth Conboy, Avneesh Singh, Leonard Rosenthol, Chris Maden, Peter Krautzberger, Laurent Le Meur, Brady Duga, Bill Kasdorf, Hadrien Gardeur, Fred Chasen, Bill McCoy, Matt Garrish, George Kerscher, Charles LaPierre


Guests: Shane McCarron

Chair: Tzviya Siegman

Scribe(s): Peter Krautzberger


Wolfgang Schindler: *I have no audio connection today (as I am not at my office) so that I may only act via IRC ;)

Wolfgang Schindler: *Wolfgang waves back @dauwhe :)

Mateus Teixeira: mateus-teixeira has joined #pwg

Wolfgang Schindler: *is it possible to ask sth via IRC only?

Wolfgang Schindler: *thx, tzviya :)

Wolfgang Schindler: *bots have a quite restricted vocab.

Peter Krautzberger: I can scribe

Leonard Rosenthol: thanks Peter!

Leonard Rosenthol: (if you need help, let me know…)

Tzviya Siegman:

Tzviya Siegman: approval of previous minutes
… => approved.

Harriett Green: new to the group, representing University of Illinois.

1. Testing

Wolfgang Schindler: welcome, harriett!

Harriett Green: digital publishing initiative

Harriett Green: thank you!

Tzviya Siegman: shane from Spec-Ops is here.
… we would like your input on testing etc.
… overview:
… WG was just chartered
… somewhat special since we carry on items from IDPF merger
… four deliverables:
… web publications and portable wp might end up be one specification with a layering
… epub4 is next epub version

Leonard Rosenthol: <- as an Illini grad, go @dauwhe!

Tzviya Siegman: dpub-aria is a continuation of the current spec
… all specs need testing, we need testable specs

Heather Flanagan: Yay for iterative testing!

Tzviya Siegman: given the amount of specs, we have a lot of work to do.

shane: daunting tasks

Tzviya Siegman: we’re not necessarily talking about browser implementations
… e.g., Readium team largely here, will feed into testing via Readium. not expecting Firefox etc to implement anything.
… we we need to make sure that we clarify epub notion of reading system and user agent.
… if reading system uses the same kernel as a browser but does something special, that might count as a different agent.

Tzviya Siegman: rk

Leonard Rosenthol: each of our documents have different testing requirements.

Ric Wright: from readium foundation perspective, there are several different variants
… SDK uses different engines
… chrome app, end of life this year
… readium cloud reader => whatever browser
… all leveraging a browser engine underneath.

shane: web engine for presentation and user interface?

Ric Wright: yes, and scripting.

Tzviya Siegman: to recap, somewhat different than usual W3C testing.
… note: FPWD targeted before TPAC.

Leonard Rosenthol: MN in winter - +1 to that :)

shane: we have a lot of work on testing.
… focused on accelerating
… breaking the problems down a bit:
… dpub-aria space is well understood, reasonable testing strategy that seems to work
… underlying a11y platforms is also well understood.
… feel free to take advantage of the smart people around that.
… re epub, lots of platforms consume epub
… at w3c, we want to be inclusive of anyone that implement our specs
… how has epub testing been done historically?

Leonard Rosenthol: (yup - no hurry)

Tzviya Siegman: IDPF had different implementation requirements, more “how do you plan to implement”

Laurent Le Meur: to tzviya, ok

Dave Cramer:

Garth Conboy: there is an epub test suite, maybe somewhat stale, tackling rendering and processing

Tzviya Siegman: tests reading systems - very manual

Garth Conboy: epubcheck tool: checks epubs against specs, also a little stale but efforts to update underway

Tzviya Siegman: is test suite for epub files - overhaul in progress

Leonard Rosenthol: from F2F, can you talk about W3C testing authoring vs consumption?

shane: authoring is not tested.

Chris Maden: Is that true? HTML and CSS validators?

shane: W3C publishes guidelines for a11y, i.e., WCAG

Laurent Le Meur: to George, the US number is USA: +1 (312) 757-3129

shane: that’s requirement on content but IIRC the only requirement.

Chris Maden: The XML specs are content specifications.

Leonard Rosenthol: how does this apply to HTML and CSS validation?

shane: we only provide tools for authors to check their content.

Leonard Rosenthol: Thanks @shaneM

Tzviya Siegman: that’s another difference from IDPF, historically provided validators, i.e., epubcheck.
… its success is not due to IDPF requiring it but people selling epub requiring it.
… to come back to testing specs.
… that’s not something that’s been done historically at IDPF / for epub
… it’s a pretty manual process

Wolfgang Schindler: I think a validator is decisive for a publishing format

Tzviya Siegman: nothing automated about it.

shane: is it correct to say: epub3 is (or will be in epub4) a superset of HTML and CSS.

Garth Conboy: both sub and superset.
… this might be less true as we define epub4.
… likelihood that we will leave more alone than in the past.

Ric Wright: Like SVG animation…

Tzviya Siegman: for web publications, while we haven’t figured details out, it’s modeled on top of web app manifest.
… i.e., a superset.

George Kerscher: Tzviya sounds good

shane: there are already rich tools for testing html/css/js/dom implementations.


shane: webplatform tests are a large collection of tools.
… these can be used to assess epub readers automatically
… if an epub reader is just something that supports the format and uses the webplatform
… then it’s more about extending tests within this framework.
… this is the way forward in the W3C
… for format, it’s mainly processing structure, not throwing errors
… whether manual or automatic might be open to discussion.
… manual is onerous but is ok and can be necessary
… some automated via format checks (manifests => associated files => feed them into existing processors)
… on epub side, we can use the tests you have.
… essentially, epub is a profile of the web platform.
… so you can define that profile and define tests that implementor has to pass in addition to web platform tests

rick: from F2F, do we need to test aspects of specs that are not part of the packaging structure, i.e., do we need to run html tests, do we need to create epubs that encapsulate all the web platform tests?
… or just those that are part of the packaging or specs we produce.

shane: I suspect it’s up to the group.

Peter Krautzberger: see what I did there, shane?

shane: depends on the architecture of the reader
… e.g., Readium built on platform browsers who already run these tests.
… would find approval in W3C to state that you use browsers that are known to run all tests

rick: went through test suite, some don’t use readium at all, just testing underlying browser engine (e.g., fetching a file and rendering)

Dave Cramer: for testing browser features, the history of epub shows that we have challenges. every reading systems has mangled some CSS, so we do need to test.

Leonard Rosenthol: also, security.

Tzviya Siegman: shane, any recommendations to get started with these tests?
… not many people have experience with W3C test

shane: start with assertions, testable statements.
… people tend to start with a wiki to capture them
… then work out how that might be tested
… that usually reveals how this could be tested
… once you have some momentum there, then work on a general strategy
… at that point bring in somebody like shane
… e.g., fetching requests, there are tests, including lots of edge cases.

Tzviya Siegman: any suggestions?

shane: e.g., benjamin.
… also browser vendor people are available to W3C.
… I will do some reflection and come back with suggestions.
… it’s a social critical task.
… for us

Tzviya Siegman: ok, we will follow up.

Garth Conboy:

Garth Conboy: see link.

2. Web packaging update

Garth Conboy: briefly discussed at F2F
… the new web packaging effort seems to be alive
… brady and garth met with some of Googlers
… they are open to working with us
… to see how it can fit our needs as well
… it is being bifurcated, formatting and signing taking to IETF while browser and packaging will remain in W3C.

Heather Flanagan: :-)

Heather Flanagan: Which working group?

Garth Conboy: IETF in Prague this month, the work will be presented there.
… after that we might get an update from them
… might have news late this month.

Leonard Rosenthol: @heatherF - I’m wondering the same thing because AFAIK it’s not be taken up by anyone…

Bill McCoy: re coordination with web platform WG
… latest conception of spec in recharter of web platform WG has drifted quite a bit
… connection with web app manifest
… we might need further coordination on that that’s broader than publishing
… not clear if co-chairs have signed off on moving to IETF

Leonard Rosenthol: spoke to our IETF reps, IETF does not seem to have taken this up

Heather Flanagan: what group is supposed to be discussing this?

Leonard Rosenthol: that’s the question. couldn’t find a group.
… maybe we can ask Jeffrey what IETF group is taking this up.

Garth Conboy: will do.

Tzviya Siegman: topic manifest

3. Manifest

Dave Cramer: one of our issues has 72 comments.
… struggling to focus some of the conversation.

Tzviya Siegman:

Dave Cramer: some premature discussion on serialization.
… somewhat worried about re-inventing the wheel.
… e.g. navigation
… web publication and the web works, we should build on that
… instead of stuff that’s only implemented in a very narrow world.

Tzviya Siegman: agreed. some issues are getting into very technical details.
… but we need to focus on broad FPWD not specifics if json is best
… we need to get the broad scope of that down.

Dave Cramer: I think we are wrestling with some fundamental issues
… trying to write something down in the broadest way possible, forces us to work through these issues.
… they will be very impactful later on.
… . e.g., idea of dauwhe going through issues, splitting off smaller bites
… that seems like a reasonable next step.

Tzviya Siegman: and don’t hesitate to close issues.

Garth Conboy: issue with high-level stuff in manifest
… if we break them down, do we risk rat-holing on the technical issues
… as opposed to driving towards agreement on the 5/6 items that should end up in manifest.

Dave Cramer: makes sense. Maybe try to move that issue more towards on consensus on the big picture.
… on required vs nice to have bits.

tim: re spawning more issues. providing a venue for discussion details like serialization.
… if we break it off now, it can continue for as long as it needs to, say a year.
… you don’t want your main issue to go on that long.
… but those one’s can.

Tzviya Siegman: note that we don’t have to resolve issues before starting to write.
… dauwhe, maybe getting something down on pixels will be better to make progress.

Dave Cramer: agreed.

Leonard Rosenthol: some of the long threads came to good points of agreements.
… e.g., what is (not) required for manifest.
… I think there’s good work already.
… happy to help sorting out spin-off issues

Tzviya Siegman: any other groups want to get started?

4. Accessibility TF

Avneesh Singh: send out call of participation

Garth Conboy: Doc of tasks from NYC:

Avneesh Singh: about a dozen interests, mostly US.
… we can only formulate our plan once we talk to wcag.
… we’ve reached out, some delays due to vacations.
… but we need to prepare well.
… once we do that, we need to figure out the timeline and sort out what goes where (on which WG )

Tzviya Siegman: rich from WCAG reached out on interest in co-chairing personalization TF

Avneesh Singh: we have to work out priorities. There are groups where we need to participate vs we need to drive.

Tzviya Siegman: agreed but they are actively looking for a co-chair.

Charles LaPierre: tentatively interested.