Publishing Working Group F2F Day 1 — Minutes

Date: 2019-09-16

See also the Agenda and the IRC Log


Present: Charles LaPierre, Luc Audrain, Romain Deltour, daisy, Wendy Reid, Jun Gamou, Laurent Le Meur, Juan Corona, Gregorio Pellegrino, Brady Duga, George Kerscher, Yu-Wei Chang (Yanni), Rachel Comerford, Avneesh Singh, Matt Garrish, Marisa DeMeglio, Benjamin Young, Dave Cramer, Daihei Shiohama, Reinaldo Ferraz, Ralph Swick, Kaleeg Hainsworth, Bobby Tung, Garth Conboy, Toshiaki Koike, Makoto Murata

Regrets: Ivan Herman

Guests: Christine Runnegar, ayukiyoshida, ryokuroda, ponyo

Chair: Wendy Reid, Garth Conboy

Scribe(s): Marisa DeMeglio, Rachel Comerford, Dave Cramer, Brady Duga, Charles LaPierre


Wendy Reid: going to CR for audiobooks, pub-manifest, sync media update
… meeting with aria and mathML groups too

Laurent Le Meur: at dev meetup tonight, we’ll demonstrate comics on readium at 6pm. location TBA.

1. lightweight packaging format

Laurent Le Meur:

Laurent Le Meur: draft note is available. shows how to put the pieces together and transmit them. small spec with specific use cases. main use cases explained in the doc.
… provide the finalized packaged publication from the publisher to the conversion house. useful for audio books going between the publisher and audio studio
… can also be used to download packaged publication from the distribution point to the user agent
… the spec goes over some very simple terminology, originating from epub 3.2 or web pub document

Luc Audrain: W3C Developer Meetup – Argos D, 1F, 18:00 – 20:30

Laurent Le Meur: based on zip
… viewed as an iso standard
… we defined a file and dir structure for what ppl have to put inside the zip. it’s totally compatible with the packaging of the pub manifest.
… we have specified a standard name for the manifest (‘publication.json’). an html entry page, if present, has to be called ‘index.html’
… manifest could be embedded inside primary entry page. only one of manifest or entry page are mandatory.
… these two core files must not be encrypted
… the zip also contains all the resources of the publication, and they may live in any internal folder of the zip
… resources may be compressed
… note that streaming compressed audio files is more complex and we recommend to not compress them
… there is an algorithm to extract the manifest from this file, taken from web pub document and simplified
… UA conformance section: UA must be capable of exploiting the archive format and opening and processing the manifest
… placeholder for examples
… the file extension is .lpf and the mime type is application/lpf+zip
… right now, each bookseller has its own requirements for how creators should submit their content. this should solve this problem.

Dave Cramer: UA conformance section - can these statements naturally lead to tests? epub spec is not successful in this regard.
… to support LPF, all you have to do is be able to access the files i the zip and present them to the next step in the processing chain

Romain Deltour: +1, “is capable of exploiting the archive format and file structure” sound too vague to be testable?

Dave Cramer: is this only a subset of a web publication user agent

Laurent Le Meur: if you are a UA capable of supporting web pub, you don’t need to open a zip
… e.g. readium can open a zip and extract the manifest and expose the different resources as an internal web pub
… the two first steps may not be among the capabilities of a web pub UA
… bullet points could be more precise in referring to the previous sections in the same document

Dave Cramer: current structure is hard to write tests for

Romain Deltour: is it backwards compatible with the packing spec in epub?

Laurent Le Meur: yes totally compatible
… we could have zip with an epub and then add the manifest. then you’ll have an LPF.

Marisa DeMeglio: .. it would have to have the mime type file at the start

Romain Deltour: this packaging format may not work for all kinds of pubs - e.g. resources might be remote
… is there a way to document which part of the target publications do not work?

Laurent Le Meur: this is not a way to “freeze” a publication. it’s a way to make an epub-lite and expose it on the web later. no link to https.
… if we want to fetch resources via hyperlink, you can still put the ? in there
… it’s intended to be used before the publication is visible on the web, i don’t see the point

Romain Deltour: it says LPF is used to exchange in-progress publications
… web pub is given as an example of a publication, but not all web publications can be packaged with LPF. there’s a venn diagram here.

Laurent Le Meur: yes true
… does not cover the need to take any web pub and freeze it

Romain Deltour: this could be stated more clearly

Dave Cramer: there’s a statement: “a package must include all resources within the bounds of the publication… “
… i wonder if we should add a note addressing for example are we explicitly forbidding things like web fonts

Dave Cramer: is there some way to clarify that, like epub, there are certain exceptions for what can go in the package in general
… epub tends to provide more explicit lists of what types of things should be in the package

Laurent Le Meur: we should review the epub spec and extract some wording from there

Luc Audrain: +1 to dauwhe re testing

Romain Deltour: Resources Location in EPUB:

Luc Audrain: testing must be very clear and this doc must be testable.
… epub 3.2 work can help here

Murata Makoto: Here I would like to say that I’m sympathetic to Dave about non-testable requirements. IMHO, non-testable requirements are nothing but dreams.

Brady Duga: to go back to the web font topic, we initially said everything has to be in the package, and then we started making exceptions
… this was not planned nor architecturally sound. it was an escape route from the corner we’d backed ourselves into.
… go back to beginning: does everything have to go in the package? is a better approach than just poking holes and making exceptions

Wendy Reid: if the resource has changed, it’s the fault of our relationship with the web
… it’s ok to ref the web with the caveat that what you referenced may not be what you expected

Charles LaPierre: can lightweight packaging format be used offline? e.g. on a USB stick in an offline environment?
… are we poking holes and creating a requirement to be online?

Wendy Reid: getting dangerously close to best practice territory
… responsibility lies with content creator

Dave Cramer: in the case of fonts, the web knows what to do if you can’t find a font

Matt Garrish: a lot of this are prob things that we should discuss in regards to audiobooks
… this spec may not be where we solve these questions

George Kerscher: Because testing is so important, is it possible to incorporate the tests into the spec instead of having testing being derived after the fact?

Wendy Reid: many audiobooks do reference resources on the web due to sheer size and streaming feature

George Kerscher: i’m wondering about dauwhe’s testing questions - can we figure out which tests we would be looking at as the spec develops, rather than writing the spec first and coming up with tests later?

Romain Deltour: +1 to mattg re not solving these things in this particular spec
… idea to have a manifest property that indicates whether it is packagable or not. might be awkward though.
… i think we should loosen the current restriction

Laurent Le Meur: we can loosen the spec - we specify that URLs in reading order must be relative

Murata Makoto: I would like to drop User Agent Conformance.

Laurent Le Meur: we can say if we see an absolute URL, we won’t require it to be in the package

Murata Makoto: i’ve been involved in OPC (open xml package format)
… conformance requirements in applications are difficult and dangerous to specify
… we should only provide data conformance
… if we are going to spend a lot of time creating test cases and methodology, we should stay away from UA conformance. it’s a waste of time.
… at least 3 formats have tried this and failed

Dave Cramer: +1 to this idea
… it seems impossible to test, and i’m not sure what testing gives us in this case. if the application cannot process this format, then it is not conformant.
… aka how would you write a bad implementation of this spec - what would it do wrong

Laurent Le Meur: if you don’t get the manifest in any situation…

Dave Cramer: then you can’t do any further processing and you’re stuck
… if you want to work with this data - we’ve described the format in enough data, so have at it, devs

Laurent Le Meur: the UA conformance here is paraphrasing the description of the format in other sections. not so useful.

Romain Deltour: one thing we can test is obtaining the manifest, because there is an algo for it

Matt Garrish: not to make light of testing, if it’s just a note, there’s no TR process

Murata Makoto: Then, the shorter, the better.

Wendy Reid: i will write a proposal to remove UA conformance section

Luc Audrain: i wonder if there will be an AudioBookCheck (a la epubcheck) - then we will need some conformance verification for the audiobook spec and the packaging spec
… some tests would have to be written

Romain Deltour: there are two things here - UA conformance and content conformance. we’re talking about eliminating the former, not the latter.

Dave Cramer: packaging part should be simple and difficult to screw up
… web world less interested in validating things; rather, let’s define what happens when things go wrong

Proposed resolution: Remove the User Agent Conformance section from LPF. (Wendy Reid)

Rachel Comerford: +1

Wendy Reid: +1

Marisa DeMeglio: +1

Toshiaki Koike: +1

Jun Gamou: +1

Brady Duga: +1

Murata Makoto: +1

Dave Cramer: +1

Matt Garrish: +1

Gregorio Pellegrino: +1

Laurent Le Meur: 0

Benjamin Young: +1

Romain Deltour: +1

Resolution #1: Remove the User Agent Conformance section from LPF.

Juan Corona: +1

Charles LaPierre: 0

Luc Audrain: +1

Wendy Reid: laurent, are there any open issues or questions you need answers for?

Laurent Le Meur: no, just waiting for the finalization of the Publication Manifest document to get a link to it

2. horizontal reviews

Murata Makoto: Deflate only? ZIP allows more efficient ones.

Laurent Le Meur: deflate only, in order to stay simple.

Wendy Reid: PING joining us. also i18n and a11y reviews happening. … i18n mostly complete (thanks ivan! )
… the only outstanding question is the potential for multilingual text in the manifest
… ivan_’s resolution is that we prob require html markup to do so

Matt Garrish: one open issue under discussion w. addison and richard is whether or not valid/well-formed lang tags are required in all the places where we reference BCP-47 … a well-formed one only has to conform to the ABNF … a valid one also has to have registered subtags

Murata Makoto: Check against the BNF. That’s enough, I would say.

Matt Garrish: the issue i had initially with it is that if it’s well-formed then you can use whatever you want, so long as it is 2-8 chars (e.g. looks roughly like a lang tag)
… i18n group says well-formedness is what matters, not so much validity

Dave Cramer: i think well-formedness is fine. restricting authors without a really good reason is a bad idea.

Murata Makoto: +1 to dauwhe
… ppl don’t specify strange language tags. they use standard ones. if they choose otherwise, ok, but it’s not common.

Matt Garrish: i tend to agree that well-formedness is fine. should we say that the first part of the language tag should be valid, or do we not care at all?

Charles LaPierre: do checkers need to be aware?

Murata Makoto: tts engine supports limited set of language tags, even if a tag is well-formed, there’s no guarantee that it’s supported

Marisa DeMeglio: we are getting into best practice territory

Dave Cramer: +1 marisa

Benjamin Young: do we need a resolution on the BCP-47 question?

Wendy Reid: no, we can just leave it be, go write your book in klingon

Avneesh Singh: thurs we will find if there are issues with a11y horizontal review. i have made a suggestion to the html files re issue #13.

Benjamin Young: zip supports all kinds of compression algos - not just deflate
… we should be clear about what we support. likely that only supporting deflate will frustrate someone. we should be up front about why.

Murata Makoto: if we use a better algo, it could be 30% smaller. deflate has a big advantage though.

Laurent Le Meur: deflate was chosen for epub
… even if zip libraries offer a choice, end user tools usually don’t
… we won’t deflate binary content. just the text.
… the 30% size reduction is not worth the complexity

Benjamin Young: why aren’t we compressing media files?

Laurent Le Meur: we copied that part from epub
… when you want to stream from the zip, deflate makes it complex

Ralph Swick: [Ralph arrives (pseudo-Ivan), with apologies for being tardy]

Benjamin Young: if we make these choices to suit end user tools, then what i produce with an end user tool should give me an LPF

Laurent Le Meur: use of deflate is a should

Dave Cramer: policing the zip details didn’t help the epub world
… we should let the libraries do what they do

Luc Audrain: +1 to Dave

Wendy Reid: PING coming at 11 for horizontal review. anything from the PING survey to discuss ahead of time?

Brady Duga: someone added privacy + security sections. there were one or two comments to maybe delete.
… we have no APIs or protocols
… should be pretty straightforward

Ralph Swick: duga is mostly right
… suspect PING will have looked at what metadata we’re requiring in the manifest
… we’re prob in good shape
… they will want to make sure we’re not requiring things that will be misused

Wendy Reid: a few administrative things

Ralph Swick: welcome a new WG member Kaleeg Hainsworth from BrightWing media

Brady Duga: PING questionnaire asks how we handle data, prevent unauth access, if we track ppl, how much user data we expose, how much metadata do we include
… many q’s are not applicable to us as we have no APIs or protocols
… anyone who allows access to this data has to be careful re how it’s exposed, but that’s not a concern for our spec
… we could add a comment in our privacy section that says we could potentially id and track a user who has downloaded files
… we have a very limited set of data that’s required, most of it is optional
… we will expose things about authors like their names but that’s not extraordinary
… security implications of using json/html/css/jpegs/pngs but we don’t have control over that

Wendy Reid: mattg and i added security + privacy considerations to audiobooks and pub-manifest specs

Benjamin Young: i don’t think author data is a privacy concern per se because it’s on web pages and everywhere else - it’s public info
… GDPR would not consider it private
… if and when we reintroduce discovery + apis, web publications and audio books could be used for private publications as well, but we’re not close to that

Brady Duga: author name doesn’t seem like it’s PII but there is concern that any automated system used to create these could share that information and might reveal pennames

Brady Duga: what should i do with these answers?

Wendy Reid: log as an issue on pub-manifest

Marisa DeMeglio: we are joined by Christine Runnegar from Ping

Ralph Swick: [Christine Runnegar, representing PING, joins]

Wendy Reid:

Wendy Reid: we completed the self-review last week

Brady Duga: i reviewed pub-manifest and audiobook profile
… we are not defining an API or transmission protocol so there are not so many privacy and security issues
… most of the data in a publication is publicly available anyway
… we’re fairly limited in the amt of data that we require in the files but we have a lot of optional info
… we don’t allow for any user tracking
… info in our files will have been put there intentionally by the creator
… we use existing formats and protocols - jsonld/css/html - no diff from web as a whole - and we don’t change any of the processing models
… you could ID users by the files they’ve downloaded but that’s generally true - we don’t expose a mechanism to reveal this info

Benjamin Young: we don’t have protocol so there are no concerns that come with HTTP or origins

Christine Runnegar: we would love feedback about what made sense and what didn’t
… privacy is not just concerned with what is private and what is public - it is concerned with info shared in one context being used in another context - so public info could still have privacy requirements
… dataset could be annotated with a note, e.g. “this dataset contains personal info”

Brady Duga: e.g. list of publications that a user has downloaded is extremely sensitive
… could also figure out who you are by what you’ve read
… however, we don’t have a way of exposing this info

Christine Runnegar: agree
… i would like to read your analysis of the security + privacy review and take a quick look at the spec, but it sounds like you’re prepared

Benjamin Young: i like context perspective rather than public/private
… e.g. what if i download someone’s voicemails and package it as an audiobook
… no way to tell from the json file what is private or public
… we’ve relied on other layers to take care of privacy, on top of the data
… the data itself in the json file does not contain any info about privacy/security

Christine Runnegar: generally what we’ve noticed in PING is that WGs may say that privacy is an implementation concern
… the problem is that there are good and bad implemens
… we shouldn’t leave it up to the browsers et al to be the police

Ralph Swick: q for Christine - what is the current advice to implementors
… what’s the pref of the community for giving best practice advice

Christine Runnegar: good question, don’t want to speak for the community, but my preference would be to be as clear and direct as possible, and all WGs should strive to do the best that they can for privacy, so saying for implementors that you should do it this way to avoid risks
… researchers or govt people can police implementors
… how can you make privacy choices detectable?
… classic case is fingerprinting - no way to easily tell that a server is fingerprinting a user if it’s passive
… active fingerprinting is at least detectable

Wendy Reid:

Wendy Reid:

Wendy Reid: a content creator can put any info in any of the fields
… was hard to phrase this - we wanted to suggest that if you are going to be including identifiable info (name, url), you should get explicit approval where possible
… weren’t sure if that suggestion oversteps our bounds

Christine Runnegar: I’m very happy to ask the PING group to take a look
… on the point of best practices, the WG may create a separate document
… other groups have done a group note document that covers what they’d like to see regarding this

Brady Duga: i don’t like this idea that you should get explicit consent - we should not make this decision for people
… that’s the role of laws

Benjamin Young: security tends to put it on the file format itself. in our situation, is the privacy vector on what you can record in the content
… you can’t prevent someone from putting whatever they want in text plain
… are there privacy concerns around extensible formats like json-ld?
… we can prevent privacy infringing things but then we prevent many valid use cases
… e.g. a pub manifest that points to private data is a valid use case

Benjamin Young: You have to do something to the plain text file

Christine Runnegar: I need clarification on the use cases

Benjamin Young: audiobooks limits it to audio files but manifests hits a broader number of specs
… it can be created by hand or through an automated parser

Christine Runnegar: so you’re capturing a bunch of information in a format. Anyone can put anything in it, even though it’s designed for a publication purpose. I think you’re right to draw a distinction. I hear duga in that laws have a role here.
… I think this has been very helpful for clarification
… I will take this analysis and wording back to PING

Wendy Reid: I am sending all the links etc via email

Christine Runnegar: in general, we want to be helpful - please share them. PING = W3C Privacy Interest Group. We want to improve privacy on the web

Ralph Swick: See Privacy Activity” (of which PING – the Privacy Interest Group – is a part)

3. Sync narration

Marisa DeMeglio: Can I show off the work I did?
… moving highlight will not use CSS pseudoclasses
… according to daniel CSS4 is not mature enough so w’ve left room to grow
… also in room to grow is text selectors
… also in our list in one text and one audio in a parallel set

Marisa DeMeglio:

Marisa DeMeglio: [demo]
… the text highlight synched with the audio and escapability allowed leaving the the reading which is important to accessibility

Murata Makoto: is it possible to skip to different sections?

Marisa DeMeglio: that would happen in the TOC which is out of scope for this demo

Murata Makoto: do we need escapability?

Marisa DeMeglio: We need it for structural elements

Benjamin Young: prior art from EPUB

Marisa DeMeglio: one of the issues we need to return to is how do we define what needs to be escapable
… the demo shows js implementation, playback including escapability and test feasbility
… we need to pick a mimetype

Ralph Swick: you need approval but its proceedural

Marisa DeMeglio: I want an audiobook adjacent format

Charles LaPierre: in terms of roles in json - could you have a type like table?

Marisa DeMeglio: that’s an open issue - we haven’t defined what vocabulary has to be used
… another is incorporating into audiobooks

Ralph Swick: See Sync Media issues list

Murata Makoto: is this going to be applicable to any html document?

Marisa DeMeglio: I would think so?

Murata Makoto: do we need to coordinate with other groups then?

Ralph Swick: yes - timed text is a good start

Marisa DeMeglio: I shared with them our latest draft but they are interested in going in a different direction (sound effects).

Wendy Reid: I will edit the duration and length issue

George Kerscher: recently a discussion came up about alt text and we saw that some UAs allow for alt text to be read aloud
… we also are seeing skippability, the ability to avoid certain elements or collapse them

Avneesh Singh: we should consider the use case for synch audio for alt text

Marisa DeMeglio: this is an important oversight - we shouldn’t say text but say element so to clarify that we are pointing to a media element

Wendy Reid: so if you are including supplemental content, publishers should be encouraged to do so as html

Rachel Comerford: See Sync Media issues list

Charles LaPierre: in terms of personalization - if the source doc is changed how does that impact the rest of the functionality

Luc Audrain: as publishers we like to not have to change the content much - in one of your issues you mention web annotation selectors
… is this something that can be chosen

Benjamin Young: the current property is called text
… if the text property were changed to fragment, then this could work across various text-based formats such as text/plain and text/csv (etc). There’s also interest from Google’s Chrome team in bringing some of the Web Annotation style text selectors to HTML URLs, so there may be opportunity to provide more of those features to this spec too.

Marisa DeMeglio: please submit a PR

Benjamin Young: do your narration fils point to what they’re narrating?

Marisa DeMeglio: there’s an open issue to allow for a url

Romain Deltour: the html spec says a new metadata name should not be created if the name is for something expected to have processing requirements.

Marisa DeMeglio: I think we were putting this in the manifest
… can you open an issue

Benjamin Young: with tools like hypothesis they don’t let you record audio but you can imagine it being added and creating a more verbose web annotation

Marisa DeMeglio: open an issue?

Ralph Swick: so if someone narrates a spec, which changes…

Benjamin Young: web annotatotions give fall back selectors

Marisa DeMeglio: I’m concerned that specifying where the anchoring takes place strays into best practices

George Kerscher: when I’m reading with a screenreader it anounces the element
… if I’m escaping, it’s helpful to know what I am escaping out of

Marisa DeMeglio: we do have an open issue about role vocabulary - namely how strict it should be
… it’s a user agent concern
… you need the user agents to recognize the roles and terms, etc
… it’s an important issue

Luc Audrain: to emphasize the pointer question - the use case of dyslexia to change the presentation of text with colored syllables
… this has a huge impact on the source text
… we would like to be able to do this in the same style as synch media

Ralph Swick: there is a working group for personalization and accessibilities

Benjamin Young: as of yet I don’t know how to do a visualization without messing up the DOM
… the AOM tries to get there
… but you don’t want to mess with the AOM
… that working group would be good to connect with on this front though

Marisa DeMeglio: it feels like a user agent space concern

Murata Makoto: CSS inserts space characters
… I’m wondering if changing the spacing would impact what is being done

Marisa DeMeglio: it depends on how fine grained the change is… Adding is likely less disruptive than subtracting

Charles LaPierre: at word level it may get more complicated if you have additional spaces?

Marisa DeMeglio: this is agnostic to the level at which you highlight

Benjamin Young: DOM order is not important in synch media except to the author?

Murata Makoto: I am summoned by the I18N WG. Will be back.

Marisa DeMeglio: I’m agnostic. footnotes will matter.

George Kerscher: I will catch up in the minutes about the meeting with Amazon Kindle folks.

Dave Cramer: not yet. We are chair-free

Avneesh Singh: all of us are standing without the chair

Wendy Reid: plans have changed. we’re not meeting with Amazon.
… I may be able to arrange a small-scale informal meeting on Thursday.
… this afternoon I’d like to tackle a few things… some open issues…
… and something that wasn’t originally on the agenda. A while ago we had a strategy meeting for the publishing activity at w3c
… we can braonstorm about what we want to work on
… first we’ll do open issues

Wendy Reid:

4. Issue #62

Romain Deltour: it’s an editorial issue
… some standards use a different terminology
… we talk about raising a warning when something fails
… but other specs use fail/error
… so we should adopt the terminology used elsewhere

Wendy Reid: we can talk about it with Matt later

5. Issue #60

Wendy Reid:

Romain Deltour: there’s something about if processing does not finish, the manifest is invalid
… and we don’t say what we mean by valid

Wendy Reid: issue 60 regarding the handling of invalid values
… we say to issue warnings
… we should look more closely about what UAs should do in the case of these invalid values

Dave Cramer: At the risk of sounding like a broken record, we should make it clear for user agents, and make it testable

Wendy Reid: we should be specific… if a URL is invalid is different than if a date is invalid

Benjamin Young: to go through pub process, we need tests
… which presumes testable things
… so if we’re not clear enough we might not get out of CR

Ralph Swick: we haven’t required comprehensive test suites
… it’s up to the WG
… but for vocab, we do want to show that each term is useful, and so it must be used by two implementations

Luc Audrain: this brings back the q about audiobook check
… we have to figure out what to do with invalid values

Dave Cramer: All the people I know working for reading systems deal with invalid EPUBs every day. I’m concerned about dealing with validation.

Wendy Reid: as one of the RS people, I deal with bad EPUBs every minute
… we create behaviours for certain common errors
… we can deal with RSC-011 :)
… certain errors like mimetype or missing files are critical failures, but we had to decide based on experience
… technically we should not deal with a book with the RSC-011 but we feel we need to

Benjamin Young: issue 60 isn’t about testing, it’s about things like forcing defaults or failing on processing
… we talk about internal representations
… are we now dealing with APIs and such?

Wendy Reid: we have to steer towards a data format
… and we still have a section on processing

Benjamin Young: if we’re json-LD, and we already have processing defined–it should choke on an invalid date

Wendy Reid: we could refer to JSON-LD

6. Area 51

Wendy Reid:

Wendy Reid: additional values for RPD
… do we need more values?

Laurent Le Meur: we have a framework, we have a property, we can extend when we want
… we know a future step will introduce visual narratives, but we don’t need it yet

Dave Cramer: In the future we should not discuss this without visual aids

Wendy Reid: laurent is right; we can defer to a manga profile

Ralph Swick: this may be one place where the spec says we reserve all other values of this property for further work
… so implementations don’t crash on values that may come along

Proposed resolution: Defer #51 in favour of the future BDCoMa Profile, but Publication Manifest should state that we reserve other values for future work. (Wendy Reid)

Wendy Reid: +1

Dave Cramer: +1

Charles LaPierre: +1

Romain Deltour: +1

Brady Duga: +1

Rachel Comerford: +1

Juan Corona: 0

Gregorio Pellegrino: 0

Garth Conboy: +1

Benjamin Young: +1

Resolution #2: Defer #51 in favour of the future BDCoMa Profile, but Publication Manifest should state that we reserve other values for future work.

7. Issue #12

Wendy Reid:

Wendy Reid: this is my favorite issue!
… what if there’s a null base URL?
… in light of recent changes to the specification, we have gotten rid of the canonicalization model algorithm
… so maybe this is a non-issue

Benjamin Young: we don’t know where these json files are used
… we don’t have an origin now
… if LPF would be to go to REC, we might have to figure out how the base url is calculated
… but until this JSON file is related to some HTML document that can express a base URL, we don’t need to say anything
… it’s blank/null by default
… there are other concerns, but this issue is not an issue

Proposed resolution: Close Issue #12, the canonicalization algorithm has been removed, origin is no longer a concern for Publication Manifest (Wendy Reid)

Benjamin Young: before we vote
… the canonicalization thing has not been removed but renamed
… maybe leave that bit out
… just say it’s a json data document thingy. might not be at a URL

Ralph Swick: do you want to capture bigbluehat’s thought that this will be a concern in the future when the manifest is is included in some future transfer protocol(s)

Proposed resolution: Close Issue #12, the canonicalization algorithm has been changed, origin is no longer a concern for Publication Manifest, but should be considered for specifications concerning discovery (Wendy Reid)

Benjamin Young: +1

Wendy Reid: +1

Laurent Le Meur: +1

Gregorio Pellegrino: +1

Juan Corona: +1

Dave Cramer: +1 with an error of 1

Brady Duga: +1

Toshiaki Koike: +1

Charles LaPierre: +1

Resolution #3: Close Issue #12, the canonicalization algorithm has been changed, origin is no longer a concern for Publication Manifest, but should be considered for specifications concerning discovery

Wendy Reid: I think this one can be closed

8. Issue number nine

Wendy Reid:

Wendy Reid: this is step 11 of the canonicalization algo
… there’s not a base any more
… that step is gone
… so I think we can close

Laurent Le Meur: I agree with the resolution
… it’s also true for LPF
… which is about something in a zip and says nothing about how it can be exposed on the web where it needs a URL

Proposed resolution: Close Issue #9, recent changes to the specification around the canonicalization algorithm mean there is no longer a base. (Wendy Reid)

Dave Cramer: +9

Garth Conboy: +1

Wendy Reid: +1

Charles LaPierre: +1

Laurent Le Meur: +1

Romain Deltour: +1

Benjamin Young: +1

Brady Duga: +1

Toshiaki Koike: +1

Resolution #4: Close Issue #9, recent changes to the specification around the canonicalization algorithm mean there is no longer a base.

Wendy Reid: two issues in audiobooks
… one of them will spawn a new issue in pub manifest

9. Issue #12 in audiobooks

Wendy Reid:

Wendy Reid: there was a suggestion to include 3 things
… abridged value
… preview
… how to include a preview
… and the last one, which will get a new issue
… the isFamilyFriendly flag
… schema doesn’t have a MPAA-type rating
… there isn’t a standard
… the retailers tend to decide
… kobo ask publishers to identify, and then verify
… it depends on jurisdiction
… i think the flag is worth having
… it’s a terrible name, but schema doesn’t have a better one
… I’d like to close this issue, and then move the isFamilyFriendly to pub manifest

Proposed resolution: Close Audiobooks Issue #12, abridged and preview have been added to the specification, and isFamilyFriendly be moved to Publication Manifest (Wendy Reid)

Brady Duga: +1

Charles LaPierre: +1

Wendy Reid: +1

Romain Deltour: +1

Rachel Comerford: when we say move to publication manifest, we’re really just saying that we’re going to open an issue

Wendy Reid: yes, just opening an issue

Garth Conboy: +1 (but don’t think we should do such a thing in pbu manifest, as I think it’s intractable)

Resolution #5: Close Audiobooks Issue #12, abridged and preview have been added to the specification, and isFamilyFriendly be moved to Publication Manifest

Rachel Comerford: +1

Dave Cramer: Plus one

10. Issue #22 in audiobooks

Wendy Reid:

Ralph Swick: [“be moved to” as in “raised as an issue for discussion in”]

Wendy Reid: there’s a mention of origin in the privacy and security section of audiobook
… (quoting from spec about where resources should be same-origin)

Benjamin Young: given what we just said about pub manifest
… mentioning it here is confusing
… we should take it out, or talk about how to handle a manifest
… so we should just delete the line

Wendy Reid: I’m happy to delete

Proposed resolution: Close Audiobooks Issue #22, remove the line from the specification and defer to more qualified groups. (Wendy Reid)

Benjamin Young: +1

Charles LaPierre: +1

Wendy Reid: +1

Brady Duga: +1

Benjamin Young: i want to add that there’s something we could call out
… that’s the use of base in @context
… you can say there’s a base of all the URLs in the JSON
… we could have a section that explains how to do that
… the advantage of @base in context is that your resource list uses fewer bytes

Resolution #6: Close Audiobooks Issue #22, remove the line from the specification and defer to more qualified groups.

Action #1: edit Audiobooks standard to remove line about the origin. (Wendy Reid)

Wendy Reid: do want to discuss sync media

11. Fragmentions

Marisa DeMeglio: a way to non-desctructively reference text in an html doc
… Can talk DOM events
… Are these standard events?

Romain Deltour: Yes, there are some standard ones
… it is easy to polyfill

Benjamin Young: Back to fragmention thing
… related to the Google thing mentioned earlier
… fragmention is only JS, will never be in a standard
… The other one is designed to be a broken URL

Marisa DeMeglio: Wants bigbluehat and Tantek in the same room to duke it out
… Personally, neutral on the options

Benjamin Young: If you keep with single string fragment identifier, only a few are refertenceable from a spec
… For HTML probably want the Chrome thing

Murata Makoto: The current list of participants shows 17 attendees.

Benjamin Young: Will only show up in the scrollto section, may have spec ramifications

Marisa DeMeglio: Double hash anchors is the previous style

Benjamin Young: Big things is want somethign you can reference

Marisa DeMeglio: I am just a note, I will never be a rec
… Are there and standard DOM events that would apply

Romain Deltour: known interfaces based on Event

Marisa DeMeglio: then our own special ones like onCanEscape
… what are the standard playback DOM events?

Romain Deltour: Posted a link

Juan Corona: What about a media event?
… are you looking for anything that sounds remotely like it?

Marisa DeMeglio: Just don’t want to reinvent

Juan Corona: What are the events for an audio or video element?

Ralph Swick: See “Media events” [MDN] – might be relevant

Romain Deltour: I don’t we need start or stop events
… Mostly created for the playback moving thing
… so there would be an event for when the element changes

Juan Corona:

Romain Deltour: Playback of page showing highlight moving, and could add a magnification to show the currently spoken text

Marisa DeMeglio: Could add that to the demo
… In what part of the spec do we define DOM events?
… A lot of it could beyond playback (eg page change events)

Dave Cramer: Did Brady talk about this years ago?

Brady Duga: Not exactly, that was layout

Marisa DeMeglio: We define an event, polyfill it, an enabled page would detect that and do something cool

Marisa DeMeglio:

Marisa DeMeglio: Any takers for issue 15?
… Where do you make the association?
… We don’t know how a pub manifest will be processed

Benjamin Young: Would leave it out

Marisa DeMeglio: No, needs to be in pub manifest for audiobooks

Benjamin Young: Oh

Marisa DeMeglio: HTML pub manifest doesn’t really exist yet
… So it is really hard to solve, and we keep going in circles

Benjamin Young: Sync media is html only?

Marisa DeMeglio: No, can be an mp3

Benjamin Young: The demo was text centric, in audiobooks it is opposite

Marisa DeMeglio: Yeah, but the user experience could still be indentical

Benjamin Young: Can you get the same result using the TOC with sync media?

Marisa DeMeglio: No, don’t want sync media on TOC

Brady Duga: general agreement

Benjamin Young: Index file that comes with audiobooks is just a landing page, right?

Marisa DeMeglio: Could be, would have to check
… Do we require it in the html head always for consistency? Or leave it optional?
… Don’t know, don’t feel strongly, but need to decide
… moving on …
… Format the explainer
… Needs to look like an explainer. Could use some editing help.

Marisa DeMeglio:

Rachel Comerford: I can help

Marisa DeMeglio: Next issue, 13
… Want to leave this not strict
… Doesn’t even have to be the word role

Benjamin Young: Are you referencing an existing vocab?

Marisa DeMeglio: No, that is a minefield
… Maybe using “role” itself is dangerous
… Any complicated structure you might want to escape, or anything you might want to skip
… skipability might be ignore page numbers

Romain Deltour: The issue of not having a finite list, it enters the UX realm
… Need translations etc
… if it is set, then the UA knows what roles to expose

Marisa DeMeglio: In epub, finding the right set of terms was hard, eg how extensible should they be?
… Like language tags, don’t be too rigid

Luc Audrain: What about the term skippable?

Marisa DeMeglio: But how will they know what I am skipping?
… eg need to say skip page numbers, but not footnotes
… for escaping it is easy

Dave Cramer: It is also a little about how you are presenting the audio to the user
… There was an abandoned attempt in CSS to do that

Marisa DeMeglio: Please join the list!

Brady Duga: You need to know why its skippable, can you give it a reason instead of a role

Marisa DeMeglio: You’d run into translation problems, but the term doesn’t need to have meaning outside of the scope

Luc Audrain: Reminds me of discussion around “purpose”

Charles LaPierre: Purpose could be reused

Luc Audrain: Where is the discussion on that?

Charles LaPierre: Have gone sideways from there

Benjamin Young:

Benjamin Young: Web annotations has purpose and motivations already
… “this is why I am pointing at something”
… synced media could say “these are the terms we are using”

Marisa DeMeglio: Yeah, that’s what we need to figure out

Marisa DeMeglio:

Benjamin Young: Could steal from the existing work

Wendy Reid: Let’s break
… strategy meeting, what to work on, ideas, talk business, communities etc. on what the future would hold. Charter expires next year… we can create a new charter and do new things. could be brought to us from others but we can suggest new ideas for future work.
… ideas for next strategy meeting, so what should we include
… could be big ideas

Benjamin Young: open to ideas around web components that could become publications, iframe needs, reinvented transclusion element (toast / switch) web components show them off get developer demand then put them into HTML maybe…
… could get stuck in JS. early experiments 2 years ago single HTML document and transcluding them singularly styled included. Virtual scroll (web component / standard web component potentially),
… I would like to revisit this, maybe polyfils can also be valid implementation.

Dave Cramer: incubation / experimentation this group may not be the place to do this. this group to take existing proving things and bring them to the finish.

Rachel Comerford: roadmaps to the future this work should be distributed. escalate items from CG to WG

Benjamin Young: how to liaison these efforts , would be nice to write a proper spec, but we are not the implementers. what are the large needs when the browsers come into the room that we have a good collective set of deliverables that we need.

Juan Corona: Likes Marisa’s work, will the WG support this work like we saw with the sync media demo

Wendy Reid: there is a SyncMedia CG is also is related in the publication manifest and audiobooks. publishing CG EPUB3 CG there is room for that.
… depends on how vocal and very specific requirement for us, and the WG has to handle bigger issue of how we handle bringing books to the web.

Romain Deltour: can Publishing WG can do horizontal reviews in a publishing context. dont have a spec for this but a lot of groups are interest in us Web Components, CSS etc.

Avneesh Singh: Where do we want to go, these are all ways of getting there but what is the ultimate goal. Do we have any strong business cases.

Rachel Comerford: we need to consider epub3.2 needs to move from Final specification to a Rec Track which would move from CG to WG.

Brady Duga: +1 Rachel

Luc Audrain: we need new standards to do more types of publications. I am against having EPUB as a recomendation;
… CPU on reading devices are getting faster better etc. but we can infuse new capabilities for the web, use incubation in community groups to foster this experimentation etc. and see what we can take further in the WG.

Juan Corona: +1 laudrain, on large DOM ideas

Luc Audrain: new standards personalization for renderings new ways to display info without changing it. new work on JSONLD an metadata. standard packaging format CSS issues still for rendering content. we need to have more effort in publishing companies EDRLab to enable more work in CSS rendering for publications. Horizontal Accessibility etc.

Wendy Reid: ProRectrack for EPUB3.2. incubation is important (web publications we didn’t have incubation and proof that it would work.)
… which is why we didn’t have this incubation and is why we decided to go in the direction we did. But with EPUB we have 20 years of incubation and know what we will get. We can then go to these issues to other groups if we are moving this to publication. for CSS, packaging etc.
… we know what the problems are and what the business needs and they are not ready to move away from EPUB. I think we can move EPUB to rec, track and can web publishing, scholarly publishing, newspapers etc we can work with them.

Dave Cramer: EPUB and large documents we don’t require chunking but went that way due to limitations of a sony reader at the time.
… abstain from EPUB rec track.

kaleeg: I have been making EPUBs for over 20 years. Publishing WG is critical for me to look for W3C specs. Humanity in this world, I think we have the opportunity to take the publication (web;book sites) to take their own place in the world and a different kind of publishing model
… we have the opportunity to create our own space and new publishing model
… we can reach out and expand in populations across the planet.

Rachel Comerford: Dave mention that he splits books into separate files but we must split them because large textbooks has to be split or they can’t be loaded on these reading systems. This future that you describe sounds exciting and the incubation that will take place in the community groups sounds as though it will be fun to participate in but what is the standard that we use with user agents, browsers, and others now? We have a carefully incubated and negotiated for consensus document that is ready to go to forward to the PWG and benefit a broader audience.

Brady Duga: you can’t change the world by writing a spec, but if no one asks for it it won’t matter ,but people are asking for an EPUB spec.
… I don’t want to work on things unless there are more than 2 parties that are interested.

Ralph Swick: EPUB3 recommendation: better EPUB/PDF if there is enough interest we can do both. we shouldn’t obsess over whether which to do as if we must choose only one thing to work on. for web to get better that Dave and others to force CSS to improve and the quickest way to do this for EPUB3.3 (rec track) or we should do something further reaching.
… we should be doing more horizontal review this group has more experience and continue telling CSS to do that.
… it will be a lot of work as Dave points out,
… lets figure out Ebooks work on the web. whats better than PDF, but we have to do better.
… lets use the talents to get that. not to the exclusion of EPUB3

Dave Cramer: there are lots of things we can do we can make EPUB better, its a 1 billion dollar industry. 10 years from now there will be a lot of EPUB still around.

Kaleeg Hainsworth:
… EPUB specs not changing the world, it changed my world a w3c spec in that I have seen the w3c spec/reports specs like decloration and when I read some specs reports and the publishing WG and I saw the possibility of an EPUB could be a a 10 billion dollar industry.
… new future is possible and new directions ahead our world changing every year a new million people gain access to the web.
… w3c has that guiding light.

Wendy Reid: we have the capacity / knowledge we can bring EPUB to recomendation… and there is a growing number of folks publishing on the web main stream publishing with massive (fanFiction) communities already doing this
… having trouble tagging their content, apply metadata
… we know how self publishers do it, why are they doing it on the web, and not using EPUB.
… we need to reach out to this community, publishing we mean trade books but thats not the only publishing we are talking about.

Murata Makoto: proceedure: we are deviating the w3c process. steering committee when IDPF was absolved in W3C. we need some changes, either we give up or we ask w3c team/committee.

Charles LaPierre: 3.2 as a w3c rec is not possible without changes to the process.

Murata Makoto: we can not remove these backward compatability.
… we are expecting a new charter if Browser vendors are involved and part of the steering committee.

Avneesh Singh: 2/3 steps after this would be great. energy on this group is low. we need a strong business case. or exciting reason to move fwd. maybe within the w3c
… explore within w3c maybe see web publishing working or a strong vision to get our members excited. we need to do something different.

Laurent Le Meur: EPUB3 as a rec. do we expect more publications as a result in these communities which are not currently using EPUB now?
… better document maybe interoperability , technical difficulties due to CSS etc.

Wendy Reid: EPUB stronger epub light reader how can we open EPUB in a browser. better closer to the web,improving html for rendering, css there is potential there.
… a billion dollar industry lets make it easier for them more understandable for reading systems, closer to the web.

Murata Makoto: Three cheers to Dave

Dave Cramer: as editor of EPUB right now I want to rewrite every word especially with Reading system conformance its difficult for everyone involved… I will do this regardless of rectrack.
… there are a lot of problems on how things are organized, how we are going to work on it and what we are going to work on and the relationship between steering and WG and business group and the WG this has created difficulties between the WG/CG and BG. need clearer responsibilities and much of this is outside process there are gaps in these.

Wendy Reid: should not talk about process stuff for us

Ralph Swick: Makoto and Dave talked about that this as a bug, but they perhaps should consider it as a feature, this community is trying to push web technology that hasn’t been pushed this way before. experiment to enhance web technologies.
… w3c needs more direction from the business models and direct more of the technical work from a business perspective.
… lets keep those conversations separate
… Makoto said “W3C specs are not concerned about Longevity” we do care about the content existed.
… 0.001 feature is important “you” get to decide what is important. the whole community reviews our work, Our work will be reviewed and commented on

Luc Audrain: EPUB as a rec, why we failed with Web publications and how it will be better if we work on EPUB3 as a rec.

Wendy Reid: End of Day 1!

12. Resolutions

13. Action Items