EPUB 3 Working Group Telco — Minutes

Date: 2021-01-29

See also the Agenda and the IRC Log


Present: Ivan Herman, Gregorio Pellegrino, Avneesh Singh, Matthew Chan, LauraB7, Masakazu Kitahara, Toshiaki Koike, Wendy Reid, Tzviya Siegman, Ben Schroeter, Charles LaPierre, George Kerscher, Bill Kasdorf, Matt Garrish, Brady Duga, Hadrien Gardeur

Regrets: Dave Cramer


Chair: Wendy Reid

Scribe(s): Matthew Chan


Wendy Reid: welcome everyone, today I want to get updates on a couple things. First is the a11y TF

1. A11y taskforce report

Avneesh Singh: See Jan 14 minutes

Avneesh Singh: See Jan 28 minutes

Avneesh Singh: these are minutes of 2 meetings in which we made some important decisions
… decision for referring to latest version of WCAG
… we will always point to latest version of WCAG, with provision of referring to 2.0 for backwards compatibility
… also, US and Canada are quite aggressive with keeping up with a11y spec
… other regions not quite so much
… don’t want to discourage them from keeping up with a11y spec
… similarly, strongly rec AA, but allow A
… adapting or improving spec to keep in line with EUAA
… gpellegrino, luc, and cristina have done a mapping between EUAA and spec
… main differences are in requiring MO, and in adding a statement that DRM should not interfere with a11y
… we are in a good position to release FPWD in Feb
… will discuss with Matt re exact date
… would also highlight one Q (not for discussion now): Should we rename epub a11y to something more generic?
… a more generic name way garner more attention, but some things, like MO, are specific to epub

Ivan Herman: does it mean that you want the WG to vote on publishing FPWD now?

Avneesh Singh: not yet, still need to discuss with Matt
… will get back to the WG after

Tzviya Siegman: whether this document needs approval by APA
… not really concern about our handing of a11y, but more about how this spec touches upon areas that affect publishing in general (e.g. EUAA)

George Kerscher: APA normally does a review of the FPWD
… because of the importance of this (owing to the EUAA), maybe we can send them a version before the FPWD
… but we probably want to wait for Matt and avneeshsingh to make some of the changes we discussed recently

Avneesh Singh: to clarify, the horizontal reviews kick in on publication of the FPWD
… no need to do it before then
… don’t want to delay the timing by going to APA first

Ivan Herman: +1 to avneeshsingh

Avneesh Singh: the FPWD is just a stake in the ground, still a lot of work to do after until CR
… just a note, changing the name has strategic considerations as it make it harder to defend the epub specific parts
… last time we gave APA a request, we were delayed by a month before moving forward

George Kerscher: so this FPWD will probably be unchanged for a while, and june is an important date for EUAA, so FPWD would be the version reviewed by Europe

Avneesh Singh: once the FPWD is release, continuous changes will be published by the automated publishing system?

Wendy Reid: no, we don’t have that turned on
… but we can republish as soon as the WG approves it
… so changes could be live, within, say, a week of the WG deciding it?

Ivan Herman: the last thing about the publication
… automatic publication means that there is a specific branch for each doc, and if there is a merge to this branch, then auto publishing will kick in
… its not instant in terms of merging PR
… but if Matt makes a merge into that special branch, then auto publishing will kick in
… if we ask APA for review, they will review not only a11y, but the whole spec (2 other docs as well)
… Also, APA may decide to delegate review of the 2 other docs back to a11y TF, but formally they review the whole thing
… also, there are procedures in the sense that we have to flag issues after publishing FPWD

Wendy Reid: the plan was that we were going to publishing another draft of the other documents along side a11y FPWD
… we’ll use this timing to trigger other horizontal reviews

Charles LaPierre: don’t we also have to initiate a tag review?

Wendy Reid: yes, but we’re going to time that with getting everything published, so everything can be reviewed at once

2. Internationalization

Ivan Herman: i started on this piece, i18n guys have a self test which you are supposed to check your spec against

Ivan Herman: See I18n Self test for EPUB

Ivan Herman: here it is, with my markup and comment
… i also added background that is mostly for i18n more so than for us
… if you see anything that needs to be changed, go ahead
… there are places where i didn’t know the answer, or points of contention that will need to be discussed
… if we agree with this as is, we can mark for i18n and get them involved
… don’t want to go into all details now, but the biggest issue is probably the bidirectional text
… i think we are more or less alright, because we have the dir attribute…
… to clarify, the only thing we have to check is the package document, everything else is under the XML, CSS, etc. specs
… but we have to check our own metadata document
… there are 2 things to mention
… one, in the RS doc, there is sec 3.2, which talks about the package document
… it does pick some items like what to do with identifier, etc.
… i think we should add something about how dir is managed
… two, there is one point where there is a contradiction between what i18n requires and what I believe we should do according to the spec
… this is what happens if there is text in which dir is set, and unicode in text has also directional note
… there is remark in spec that says unicode inherent directionality takes precedent over dir
… but this exact opposite of what i18n asks for
… this is a classic example for bidirectional text, e.g. text that starts with latin character, but the text is hebrew
… this should be marked dir rtl
… but spec seems to say that unicode implicit direction takes precedence
… seems like a bug in the spec
… also don’t know what happens if title text contains unicode marker for rtl, does RS correctly handle that?
… spec should say that it should be handled
… these were the areas where i saw we had issues to discuss
… Dave had also open issues, e.g. reading direction in spine
… also, do we say that default of dir is ltr in spec?
… so please look at all of these issues, and consult the wiki above
… if we arrive at consensus on all these open issues, then I can flag them (and the wiki) for consultation with the i18n folks

Wendy Reid: I don’t think we say that we have a default directionality
… we probably use the HTML default

Brady Duga: presumably the default has to be auto?

Ivan Herman: that was the other thing, dir doesn’t have an auto setting
… not sure what auto would mean
… another issue, in HTML if you have a <a> there is hreflang attribute
… (i.e. lang of the linked resource)
… we don’t have that in the package document

Wendy Reid: any questions about i18n?
… so we haven’t submitted this to them yet

Ivan Herman: i would love it if people would review the wiki I prepared above

Brady Duga: I will look it over

Ivan Herman: other possibility is that I could raise them as separate issues without involving i18n for now, and we can discuss issue by issue

Brady Duga: separate issues are nice, easy to keep track of things

Ivan Herman: okay, i’ll create the separate issues

Wendy Reid: we have to do the other checklists for the other horizontal reviews (incl. security)

3. AOB?

Wendy Reid: okay, there are open issues in github right now
… would be good to get input
… particularly page progression direction, and conformance re. MO
… have a great weekend, see you all next week!