- From: Dael Jackson <daelcss@gmail.com>
- Date: Wed, 20 Nov 2013 07:29:29 -0500
- To: www-style@w3.org
DigiPub Joint Meeting
---------------------
- The DigiPub group met with the working group to present some of the
work they've been doing, especially on pagination and page
breaks.
- There was a request for a list of priorities broken down by market,
preferably with examples.
- Possible issues that were discussed included multi-page spreads,
headers and footers, splitting images, and internal
cross-references.
=====FULL MINUTES BELOW======
DigiPub Joint Meeting
---------------------
ScribeNick: fantasai
Markus: This interest group is charged with producing reqs and use
cases for the digital publishing industry for the open web
platform -- things digital publishing wants to do that web
platform can't do yet.
Markus: Pagination is really important, obviously.
Markus: The point of this meeting is to go over our progress, and get
feedback on it.
Markus: Our question is, how should we structure this, document this,
what do you need?
Markus: Here's our draft of pagination, which Dave Cramer (dauwhe) has
been working on.
Markus: Sits in context worth mentioning quickly.
Markus: Most of you know about JLREQ, which was worked on for 5 years
before published.
Markus: There are other similar documents within i18n starting up,
Markus: This document is going to be LatinReq,
Markus: To capture publishing requirements for Western typesetting.
Markus: It will be used to draw concrete requirements for CSS.
Fantasai: http://www.w3.org/TR/pagination/
dauwhe: ...
dauwhe: Looked at JLREQ, realized modeling it on that was impossible.
dauwhe: This is just beginning of what we hope to achieve.
dauwhe: We want to describe what print publishing has been doing for
last 100s of years.
dauwhe: And then see what is useful for digital books.
<Jirka> URL is http://w3c.github.io/dpub-pagination/
<astearns> http://w3c.github.io/dpub-pagination/index.html
dauwhe: For this document, I just started writing some introductory
material and then diving in to start describing, even in
simplest cases, what the rules are publishers working with on
books and what kinds of things do we care about.
dauwhe: That brings up a lot of the issues we need to deal with.
dauwhe: Simple example here of a fake edition of Moby Dick to
illustrate widows/orphans.
dauwhe: It brings up the concept of what exactly is a page in context
of an open web platform;
dauwhe: Concern with position of this particular line in the page,
concept of concept of spread.
dauwhe: CSS has a setting for saying how many lines it's ok to leave
at top of page, but in most UAs if you set that, what it does
is it moves a line from the previous page,
dauwhe: Which horrifies the publishing community by having two halves
of the spread not aligning.
dino: What do they want?
dauwhe: The first step is to go back earlier in that particular
chapter and make some changes that will make the problem go
away.
dauwhe: First thing we would try is making the previous spread
shorter.
dino: What do you do if it's the first page of the chapter?
dauwhe: Chapter usually starts a bit down the page, might shift it up
a little.
ChrisL: These are all manual interventions. You need rules for a
formatter.
dauwhe: Think this problem is mostly solvable, conceptually.
dauwhe: A large aspect of the labor going into producing traditional
books is solving these kinds of issues.
ChrisL: First point, if it's not solvable automatically, then CSS
can't do it either.
ChrisL: If it is solvable and CSS can do it, then this saves costs
because can be done automatically.
dauwhe: We want to follow good craft of layout.
dauwhe: If we have that automatic solution and apply that to digital
books, will be huge increase in quality of things out there.
Markus: Also, the scope of what we're trying to do with this is both
hard-copy formatting, which is a lot of the work that Dave
does, to prepare (using HTML+CSS in this case) paper copies.
Markus: But it's also dynamic pagination in reading systems, which ...
Markus: Is a huge overlap for publisher
Markus: The cost of doing these recalculations; offline it's okay if
it takes a few minutes, but in dynamic ebook environment not
so okay.
dauwhe: I don't have sense of algorithms for that.
dino: I wanted to do a lot of these in iBooks, but it's hard to do.
dauwhe: Yeah, everything prior in the document is sort of in scope, so
searching lots of possibilities.
dino: Even handling widows and orphans as we do requires backtracking,
which we don't like to do,
dino: Going back more than one page / column ... :(
SteveZ: A question comes up, is there a level of degradation that
people can live with in the e-environment that wouldn't be
acceptable in print environment, but would be good enough?
SteveZ: Good enough beats out best.
[dauwhe grumps about Kindle]
dauwhe: I have higher expectations. I don't want to give up without
trying.
dauwhe: I want to raise the standards of the digital world to match
what's possible in print.
dauwhe: Other issues ...
dauwhe: Common example in novels ... I'm calling them space breaks
<astearns> http://w3c.github.io/dpub-pagination/index.html#space-breaks-and-ornaments
dauwhe: A blank line or two where the thought changes or scene
changes.
dauwhe: At the page break, it looks like misalignment, so if it falls
at the page break you put in visible element, like asterisks.
dauwhe: So you want to know if element is at top or bottom of page, to
change its appearance. That's a whole class of things that
want to change.
dauwhe: I dream of :top-of-page pseudo-elements, e.g., to address a
whole class of issues here.
dino: I'm sure our team would love that as well.
dauwhe: As we said, we're just getting started with this effort.
dauwhe: Filling in some of these sections just to learn how to work
with this, what level of detail is appropriate.
dauwhe: Would love to know what you guys would find helpful,
dauwhe: What can we do for you, to help make this move forward.
dino: Priorities would be an important thing.
dino: For you as the industry to say what is hugely important,
dino: Especially if you have data to back that up,
dino: And break it down market by market,
dino: e.g. maybe widows isn't important in Korean.
dauwhe: Talking with others about similar issues in traditional
Chinese, e.g. don't want just one character in a column,
dauwhe: You want a minimal amount of content on a page,
dauwhe: etc.
dauwhe: Want to minimize disruptions as reader goes from one page to
another.
dauwhe: So expect lots of commonalities among writing systems.
Markus: Request was for, prioritization of all these things, and also
locale breakdown.
dino: We're often told by Japanese publishers, if you don't support
this type of annotations or whatever, it's a showstopper.
dino: It would be great to get similar types of priorities.
dino: To give another example, italics or obliques in Japanese
vertical text.
dino: Possibly it's very uncommon and therefore we don't need to spend
so much time on it.
dino: So would be helpful to know how many people are actually
impacted by this.
* Bert thinks of a slider to set the time/quality tradeoff. As
computers get faster, the search algorithm can search deeper
and find better layouts, even in real time renderers. And
authors can add metadata to indicate which aspects are most
importan to get right in a given document.
SteveZ: One other thing you hit upon is, some things the known
algorithms involve backtracking.
SteveZ: Whereas other things, like asterisks, which don't really force
backtracking.
SteveZ: So to an extent knowing tech used to implement is useful thing
to identify as well.
astearns: I don't think you need to prioritize everything in the
document against each other, but to surface the highest-
priority items.
fantasai: With regards to how much detail, more detail is always
better.
dino: Examples are great.
dino: e.g. looking at your example, I think we implemented widows
incorrectly.
[dauwhe notes that books with short dialog are hard to page, 1-line
paragraphs are great, 3-line paragraphs are really hard]
dauwhe: 3/4 of a line stranded on top of page is not so bad as half a
hyphenated word.
dauwhe: You see that in ebooks often, it's noticeably bad.
dino: Would also like to see examples of inline images. Never know
what to do there.
dino: Obviously you never want to split an image across a page, people
say that's bad.
dino: But at what point is it better to put on the page by itself,
etc.?
dsinger: Might shift it down by some lines.
fantasai: float: inline; ?
dauwhe: Something that comes up a lot in higher-ed textbooks are an
elaborate series of rules for placement of images and
ancillary text. How far can they move away from their
reference? What are best practices for stacking floats?
dauwhe: Want to document these and similar issues.
Markus: On the topic of issues to cover or not, we're discussing scope
at high level.
Markus: I want to ask this, are there things that we should focus on
or omit that would help right now?
Markus: What topics would you like us to prioritize?
SteveZ: One that's currently under discussion is footnotes and their
treatment in various columnization strategies would be useful.
SteveZ: Another I noticed is captions. Number of different strategies
for handling captions, especially when they're not around the
image.
ChrisL: Another thing would be things which are used in print but are
also used not in print, and would therefore benefit everyone.
Things that books do better but everyone would appreciate.
ChrisL: That would help generate interest for implementers not focused
on books.
dauwhe: I worked on custom publishing system for textbooks where there
were a lot of internal cross-references, e.g. "in next chapter
we find x", but there's no next chapter, what do you say?
dauwhe: Lots of interesting things to look at there.
dino: It would be interesting to get feedback on what digital book
publishing concepts translate to the world without pagination.
dino: What then do you do with footnotes?
fantasai: position: sticky!
ChrisL: One characteristic is not just that it's scrollable, but also
dynamic. You can just pop-up the reference, right there where
you're reading it, rather than at the bottom of the page where
you're not.
ChrisL: But then you need to be able to share the markup.
fantasai: Stickily-positioned elements overlap, so that wouldn't fly
for multiple footnotes.
dino: Would also help for [...]
<TabAtkins> We need to fix the overlap issue for normal stickypos
anyway.
<hober> TabAtkins: it's a feature, not a bug.
<TabAtkins> hober: Not quite. Having stickypos be able to push other
stickypos out of the way is what's needed some times, but
other times you want to stack them.
* fantasai thinks she agrees with Tab
<hober> TabAtkins, fantasai: I suspect wanting things-with-funky-
layout to avoid other things-with-funky-layout is orthogonal
to the form of funky layout.
<TabAtkins> Frex, lots of <h1>s on a page with no <section> elements,
overlapping *kinda* works, but not really - transparent
backgrounds or different heights ruin it. You want them to
push each other out of the wya.
* fantasai disagrees and agrees with Tab
<TabAtkins> But you want an <h2> to stack under the <h1>, and be
pushed away by the following <h2> or <h1>.
* sgalineau supports Ted's position:funky proposal
Bert: It would be helpful also to know fallbacks: If you cannot get
the thing X that you really want, then what are alternative
solutions that would be acceptable?
dauwhe: If I can't get a unicorn, can I get a pony?
Markus: Back to pagination
Markus: Spreads,
Markus: Showing 2 pages at same time,
Markus: Very often in textbooks, they have boxes that occupy two pages
of spread, e.g. an image.
Markus: Are spread behaviors something to go into detail on, or not?
dauwhe: ....
dauwhe: More print-specific than other things here,
dauwhe: That apply to paginated view in ebook reader.
dauwhe: On the other hand, spreads are kind of like columns, so who
knows?
SteveZ: You still have issue of figures that cross.
SteveZ: Is there much of a difference then?
Liam: If you think of a spread as a unit you're looking at at the same
time, and consider brochures,
Liam: Then you might have a 6-page spread,
Liam: So have to think a little beyond 2-page spreads.
* fantasai liam++
dauwhe: See things like that in online learning, having things
side-by-side, proves to be a useful concept.
r12a: Spread .. differences in printing material.
dino: Are you asking are footers and headers and gutters important in
ebooks? Yes.
fantasai: Well, gutters wouldn't be, but footers and headers would be.
dauwhe: Headers/footers help reader identify context within a long
work. Transcends type of media used to render the work.
* fantasai position: sticky!
[dauwhe side discussion of gutters and how to chop things like maps so
they look good when printed and bound]
SteveZ: I think you've given us a good beginning, and I strongly
encourage you to keep going.
SteveZ: Everything so far is valuable.
dauwhe: I'm in both WGs, so it's easy to talk to me. I appreciate any
comments/criticisms/whatever.
dauwhe: I look forward to really interesting arguments on what pages
are...
r12a: Any ideas on how XSL technologies map into things needed here?
r12a: Can it say anything about what CSS is doing?
Liam: Not so easy to answer...
Liam: Going forward, during paris f2f, we asked about possibility of
starting a task force to do paged media work.
Liam: There was approval for that, but expected in the next couple
weeks with regards to a task force within CSS WG,
Liam: To discuss paged media things.
Liam: The bigger question is in regards to XSL:FO, we closed that
working group.
Liam: Partly that was because we have many people interested in CSSWG
and almost no one joining XSL WG.
Liam: ...
Liam: Hopefully we will be pushing forward on areas of CSS that do
these things.
[silence]
dauwhe: So, we're eager to continue on this work and will do what we
can to help CSSWG.
dauwhe: Thank you for your time and attention.
Bert: Before we close, what's the next step? Can we have more often
joint sessions?
Bert: To keep up to date on each others progress?
dauwhe: Maybe we can work out some communication method or status
update between the groups.
Markus: ...
Markus: We can do somehting like today, but more focused on a
particular issue / area.
[dauwhe invites people to join DigiPub IG]
Received on Wednesday, 20 November 2013 12:29:57 UTC