EPUB 3 Working Group Telco — Minutes

Date: 2020-11-12

See also the Agenda and the IRC Log


Present: Shinya Takami (高見真也), Masakazu Kitahara, Matthew Chan, Toshiaki Koike, Jun’Ichi Yoshii, Ryo Kuroda, Naomi Yoshizawa (吉澤直美), Nellie McKesson, Daihei Shiohama (塩濱大平), Brady Duga


Guests: Dave Cramer, Florian Rivoal

Chair: Shinya Takami (高見真也)

Scribe(s): Dave Cramer


Shinya Takami (高見真也): welcome everyone

1. align-x-center

github: https://github.com/w3c/epub-specs/issues/1380

Shinya Takami (高見真也): this is not supported by major reading systems
… and perhaps not very many files use this?

Toshiaki Koike: proposed on github

Dave Cramer: https://github.com/w3c/epub-specs/issues/1380#issuecomment-719108924

Toshiaki Koike: every reader can implement
… this feature is supported by voyager reading system, but not many epub files use this feature
… he expects epubcheck would not mark as error

Dave Cramer: not sure how epubcheck will deal with <meta> properties… it actually might be easier
… am I correct that people don’t really see this as necessary/

Shinya Takami (高見真也): yes

Proposed resolution: deprecate rendition:align-x-center (Dave Cramer)

Nellie McKesson: +1

Matthew Chan: +1

Jun’Ichi Yoshii: 0

Ryo Kuroda: 0

Toshiaki Koike: +1

Masakazu Kitahara: +1

Shinya Takami (高見真也): +1

Daihei Shiohama (塩濱大平): +1

Resolution #1: deprecate rendition:align-x-center pending confirmation on NA call

2. CSS Writing Modes reference

github: https://github.com/w3c/epub-specs/issues/1389

Dave Cramer: this was done deliberately

Florian Rivoal: there were several kinds of changes to writing modes between its reference and rec
… some are easy to deal with; some less so
… from a behaviour change, there hasn’t been much, if we ignore the syntax
… there is one part which has changed a lot, but it’s the detailed, nuanced part
… and I doubt that epub readers implemented the correct part
… the old behaviour was badly defined
… about sizing elements with mixed elements
… automatic sizing inside horizontal text inside vertical text, for example
… I would be surprised if that was a problem
… but there have been syntax changes
… we might need to maintain a list of aliases
… and just document that
… more tricky is that some features have been removed
… hopefully they haven’t been used
… some features were moved to level 4
… text-combine-upright: digits was moved from level 3 to level 4
… and this one may still change because browsers have not yet implemented
… so if you document the syntax in the EPUB spec
… and then say for behaviour look to writing modes
… the trickier one is the complete removal

Dave Cramer: https://w3c.github.io/epub-specs/epub33/core/#sec-css-prefixed-writing-modes

Florian Rivoal: [makes same points (presumably!) in Japanese]

Dave Cramer: so we can just point to the current spec, and identify aliases to current syntax and behaviours
… I think there’s work for the editors to do to clean up this section, then we can get feedback from the wg

Shinya Takami (高見真也): that sounds good

3. Patent Policy

Shinya Takami (高見真也): w3c released a new patent policy right after this WG was chartered
… so we are officially using the old patent policy
… but we could switch to the new patent policy
… so I invited Florian (from the W3C AB) to explain the new policy

Florian Rivoal: one main difference between old and new
… with the old policy, when you reach CR, you get asked if you have any patents you want to exclude
… if you say you won’t give a license, then a special group is formed
… if you don’t answer, or say it’s ok, you promise to license them to everyone at no cost
… you don’t make this promise immediately; only when the spec becomes REC
… so there’s no license during CR; just a promise for a license later
… with the new policy, same thing–you get asked at CR
… but if you’re OK with the patents at CR, you license them immediately (rather than waiting until REC)
… another minor difference: when the old policy was written, people imagined that one WG produces one REC, then it’s over
… they did not consider groups publishing multiple or updated specifications
… so there was ambiguity

Dave Cramer: we have multiple specs. does it make sense to move to the new policy?

Florian Rivoal: yes. with multiple specs, there is potentially an issue… it would likely work out
… the old patent policy is not clear about the same spec going to rec multiple times
… especially if some people leave the group

Proposed resolution: move to the 2020 patent policy (Dave Cramer)

Nellie McKesson: +1

Matthew Chan: +1

Jun’Ichi Yoshii: +1

Shinya Takami (高見真也): +1

Toshiaki Koike: +1

Ryo Kuroda: +1

Masakazu Kitahara: +1

Daihei Shiohama (塩濱大平): +`1

Daihei Shiohama (塩濱大平): +1

Resolution #2: move to the 2020 patent policy

4. unsupported features

Shinya Takami (高見真也): deprecated features: authors are strongly recommended not to use
… reading systems may support
… legacy features: authors may include
… reading systems must not support
… I have a suggestion that we need a third category
… where authors are strongly recommended NOT to use
… reading systems should support

Dave Cramer: another factor is epubcheck

Florian Rivoal: I think I agree
… I propose that the third group be called “compatiblity features”
… maybe for epubcheck a low priority warning
… like in verbose mode

Dave Cramer: there are info alerts in EPUBCheck

Florian Rivoal: it’s also probably useful to say what happens if both properties are used

Dave Cramer: what about the cascade

Florian Rivoal: the model also tells what happens from script

Toshiaki Koike: (in Japanese)
… with current versions of epub check, when using vendor prefixes, we don’t issue any warning or info
… should we issue an info only for epub prefixes?

Florian Rivoal: my opinion is that it is low priority
… when we know what unprefixed property should be used

Dave Cramer: we’re not sure how much we want to enforce restrictions on CSS

Shinya Takami (高見真也): in conclusion, we will continue to discuss unsupported features. I will file a GitHub issue.
… and we can continue the discussion there.

RRSAgent: make logs public
… make logs public
… bye

5. Resolutions