See also: IRC log
<trackbot> Date: 13 January 2014
I can scribe today?
<lizadaly> I hear plenty of beeping
<dauwhe> Is the conference line actually working? I've heard no human voices.
<lizadaly> I just say 'hello' but heard nothing back
<lizadaly> so I guess that's a 'no'
<dauwhe> I've heard nothing.
Other than beeps
<ivan> heh neither do I
<mgylling> /me its a tibetan-style concall this week
<Luc> neither do i
<lizadaly> I assume Markus is just bored
<ivan> :-)
<lizadaly> Zakim has a case of the Mondays
<mgylling> ivan, should we restart or sth?
<lizadaly> Hear that!
<Bill_Kasdorf> I can hear now
<lizadaly> How do I tell it which line I'm on?
<Bill_Kasdorf> I mute and unmute locally. Since I can hear, do I need to unmute on Zakim?
<scribe> scribe: azaroth
<Liza> BOOM
<Bill_Kasdorf> How?
<Bill_Kasdorf> 60#
Markus: As we decided last week,
another week on LatinReq focus. We ended the last call thinking
we scratched the surface, so will continue another week
... before we do that we need to approve previous meetings
minutes
... Approved
Ivan: An admin thing before we
start, I just want to warn you that we had a security breach at
W3C.
... at some point you may be asked to change your
password.
... When you try to log in to a member restricted area, you'll
be asked to change. We're testing the system now with the
team.
fjh: When we see that, we should just go ahead and change it?
Ivan: Yes, you can postpone for two weeks, but will need to do it at some point
Markus: Dave, can you give a quick recap of what we covered last week to get back into the mode?
dauwhe: I droned on about
dropcaps for a while, some low hanging fruit for CSS. talked
about some issues people are having in ebook systems, such as
hyphenation and justification
... We talked about image and caption sizing, quite a bit about
various efforts to bring typographic sophistication to the
web.
... On Sunday I updated latinreq document, alan sent a nice
intro text on grids. Added a section about justification and
hyphenation, as a concern for a lot of us.
... so I thought today we could talk about that this week, and
about grids and underlying issues they may solve
... Just sent the list a detailed email about table
alignment.
... Anything I forgot to mention?
Markus: We had email thread from
last week where ben and luke posted some issues. Felt to me
like if we cover table alignment and so on today, that's a good
amount
... tell us about H. and Justi~n and where we should end
up?
<astearns> dauwhe: could you post latinreq URL? The task force wiki page has the wrong link.
Dave: A couple of things. I wonder if we're at the point where we or some other group need to be detailed about what we want to see. What does it mean for a book reading system to do a good job with the text
<dauwhe> http://w3c.github.io/dpub-pagination/index.html
<tzviya> http://w3c.github.io/dpub-pagination/
Dave: so one possibility on
Hypenation... a lot depends on dictionaries. Had experience
with good and bad ones. Is it possible to say which is good or
not so good with various languages?
... could contribute best practices on how to build it into
rendering engine
Bill_Kasdorf: In a lot of disciplines there are specialized dictionaries needed, due to terminology
<liam> +1
Bill_Kasdorf: some way to accommodate this is important, esp for medical.
Dave: At individual title level,
exception dict is important
... per file or copy, the dictionary is a critical piece to
supply hyphenation exceptions
Markus: To understand, the document provider should declare which dict to use?
Bill: In addition to the built
in, should use Steadman's medical dictionary as authority for
hypen~n
... sometimes a publisher will have an individual list of words
it knows will be problematic in a title and provide list of
dozen such terms
Markus: So not in a dictionary, local to the particular title
Dave: various systems have ways to enter new terms to override
Markus: Sounds like 2 clear requirements to formulate
Tzviuya: Same as BIl, but as we get into more advanced fields, need expanded on. Medicine is broad, but grad level chemistry things get complicated
<Zakim> Liam, you wanted to note it may vary within a document
Liam: Live in a multilingual web,
and multidiscipline. Might need multiple dictionaries for a
single document
... saw this with print publications in 80s
Markus: Any effort in W3C in this area so far?
Liam: Nothing beyond what we did in XSL FO.
Dave: Is there a standard for even describing hypen~n like TeX style that aren't easy to understand
Liam: Not aware of any
... eg words in German that change spelling when you hypenate
them
... might be in OpenOffice
Luc: Proper way to solve this is
perhaps to rely on hypen~n inside the text
... in a limited environment could be possible to prepare text
with the special words / preferred hypenation
<astearns> so far, all CSS says about hyphenation dictionaries is that the user agent must choose a language appropriate resource
Luc: question may be to find a
good composition engine that takes into account the preferred
places in the words
... as in EPUB with fixed content document, could be prepared
in advance
... so not to rely on external dictionaries
<Zakim> Liam, you wanted to note trend is away from inline resources and towards separate resources; also, browsers today *do* do hyphenation, just don't support per-page or per-site
<astearns> +1 to Liam's point
Liam: Trend is away from inline
and towards external, like CSS and fonts
... when people look at 2 or more pages, is better
... but don't support hyphenation. Search in page stops working
etc.
... More that needs to be specified
Bill: Almost always a dict is
used in tandem with an algorithm as it doesn't provide all
possible words
... if present in dict it takes precedence
<Luc> +1
Dave: Using dict in a very broad sense
<astearns> dauwhe: you can encode prioritization in some systems (two dischys for a better break, for example)
Bill: In all languages many examples of limitations where you can't do it without context, eg pres/ent and pre/sent
Dave: Situation out there in the wild is bad, we want to improve it, but not going to get perfection
Bill: Yes, not attainable
Dave: There's a lot we can do in
this area for requirements
... perhaps moving on to justification
Luc: Main question remains, if
dictionaries used for exceptions, on algorithms used in OWP to
calculate line length and breaking
... or declare them?
... Don't know which should be used but would like to know
which are available? Does each browser have their own?
<liam> [document providers will be able to say they prefer a particular algorithm from a well-known list; I hope to see at least two supported (first-fit as now, and n-line) ]
Dave: It becomes a bit intertwined with justification at that point to decide
<Zakim> astearns, you wanted to mention ladders
<azaroth_> Markus: Okay, moving on...
<azaroth_> Dave: Should we talk about tables a bit?
<azaroth_> ... I think this a good and interesting topic
<ivan> mail sent just now
<azaroth_> ... Ben has done a tremendous amount of research into automated rendering for HE texts
<azaroth_> Ben: I sent a mail to the list. Higher level -- line on character support, but can be on any character often decimals
<azaroth_> Second is support for flush left center. Not sure there's a standard term.
<azaroth_> ... FLC is a common variation on FL and FC, not supported in things even like InDesign
<azaroth_> Ivan: What does it do?
<azaroth_> Ben: What's the difference? FLC says look at the column, find the longest line in the entries, center that line, and then take all the other lines and flush left on that longest, centered line
<Bill_Kasdorf> That is used in poetry as well as tables
<azaroth_> Markus: Flush meaning align the left most character with the base
<azaroth_> Ben: Yup, align left
<azaroth_> Bill: Also done in poetry
<azaroth_> Ben: Third is preventing line wrap if you have cells aligning on a character or word
<mgylling> important point re poetry, means this is not table-specific alignment behavior
<azaroth_> ... Fourth, table widths. Lots of notes about requirements. Ask group if I should document?
<azaroth_> ... idea is that tables don't size to some arbitrary dimension, set by fixed width
<azaroth_> ... idea partially aesthetics and relationships between tables
<azaroth_> ... 4 tables that showed similar data and should be considered together
<azaroth_> ... if rendered at diff sizes may be hard to see they're related
Dave: can't preset the width, so
have to be able to set table to some choice of width that is
provided
... those are the high level points so far
... Alan? you replied on list, can you read it back to
discuss?
Alan: You had steps for aligning
on a character eg decimal. didn't seem right
... centering portion can't be part of the algorithm
sometimes
Ben: Circumstance?
Alan: Gave some sample data in
reply.
... Can result in case where nothing is centred and that's
appropriate
Ben: Yes, true.
\me notes scribe isn't understanding discussion
Alan: Focus more on discussion and less on an algorithm. Could be a faster algorithm. Just talk about requirements not necessarily steps
Ben: Fair enough. I think in this
case I think we want to stay away from implementation
details
... trying to give examples that there's additional details to
consider
... agree that when more digits on one side in one cell, and
more on the other side in another cell, but no individual cell
would be centered
Markus: Shall we go through Q?
Tzviya: Snapping to width is
relevant, often related to trim size. Data relevant to a
section needs to be presented in the same way.
... also spans and straddles -- can be difficult to achive with
CSS, esp with centre the content within a span. In HE content
often see vertical text, not sure if we want to mention in high
level reqs but will come up
Ben: Agreed
<Bill_Kasdorf> Do you mean vertical text or rotated text?
<astearns> I think we should definitely mention Latin vertical text requirements
Liza: To what extent do we want to consider tables in mobile devices where there's no room
<tzviya> +100
Liza: problem with basic html in reading systems
+100 from me too
scribe: Do we want to dictate what the behaviour should be when the reading system can't fill the space
Markus: Something that Tzviya has talked about as well. The most basic things not working. A separate cluster of requirements?
Dave: First thought is it sounds
like a separate cluster of fallbacks for tables in limited
screen real estate, how tables should degrade. Haven;t thought
about it a lot
... expect other people have spent more time
Markus: What task force? Tzviya, under adaptive?
Tzviya: I mentioned it. If we
talk about tables in this document, we should mention it. That
it won't render everywhere. Also Bill was considering -- not
everything can function as a table in mobile
... not sure a fallback would be the way we define it
Dave: Just using to mean some simplified arrangement
Liza: Not to increase the scope, but need a position on it as a real issue
Markus: Agreed, been up before and we need to cover it
Dave: latinreq might be the place to describe the high end use case, what we want to do with tables, but not the place to work on more restrictive circumstances
Markus: Natural home? Tzviya?
Tzviya: Still trying to figure
out what my area is :)
... do have a use case, so can bring it up. Sure.
Bill: Just to point out that
there's an obvious behaviour that compensates -- the ability to
hide rows or columns. Typically not reading everything, just
looking something up
... don't care about columns 2-4 to look up something in column
5
... so if you can hide them, you can see it on your phone
Tzviya: Don't necessarily agree. Sometimes need everything to do a comparison.
Bill: If it's a table, then given the ability to look at portions is useful. Not a complete solution
Markus: Where were we with alignment in columns
Ben: I'll clean up the language to correct hte language. Screenshots ... how to provide?
Dave: I can put them in the doc
if you email them to me
... next comment was can we have some illustrations
Markus: Sounds like we had unanimous agreement was requirements, not to mandate algorithms
Dave: Some sort of tricky example to align on character, describing these kind of arrangements
<Zakim> astearns, you wanted to comment on algorithms
Dave: in scope for this doc
Alan: Algorithm the computer has to go through seems out of scope. Manual algorithm for print production, those steps are very useful to describe
Dave: Yes, this is how we think about it, the concepts we apply to get the desired end result. Agree those are good
Markus: Dave, you also mentioned grids as a topic
<dauwhe> http://alistapart.com
Dave: Thanks to Alan for the
text. Provided a lovely example of the problem
... Something I mentioned last week, the idea of the grid seems
important, also trying to get at the underlying principles.
Alignment goals that grids implement
... trying to get to a more fundamental aspect of the problem.
Looking at the example there, what is supposed to align with
what, that would make it look less horrible
... at bottom, there's a section more from alistapart, 3
articles with varying structures so nothing aligns
... main text isn't on the same grid, varying numbers of lines.
all over the place
Alan: One of the things I'm
planning to do is put some screenshots together to show before
and after, what controls you'd need to achieve the after
effect
... been considering for grids are theoretical, only work on
the tools. for ppl using baseline grids, they have some manual
steps, would be good to get feedback for requirements
Dave: Love to see some examples
of grids in use. Can point out that grid enables this to work.
Underlying design principles to codify
... what grids need to achieve
Bill: Makes me think of related
use case- ability to set a vertical marker that other things
are aligned with
... not a pre existing grid, just an alignment point. On the
list?
Dave: Applies to lots of things. Side by side text, side note.
Bill: translations for example
Dave: Yeah
... way to go is if you have examples, then send them along
Bill: Yes not a grid, just made me think of it
Dave: Larger issue of alignment.
How does this object relate to that object
... whole subject has lots of diff manifestations. As many
examples as possible to let us come up with what we need to
do
Markus: Bill, I thought you were going to talk about alignment in 2 page spread. points sometimes cross page boundaries.
Bill: That's true. More universal
than that. DOn't want people to focus on page based rendering,
relevant in reflowable
... good question though, would the same mechanism work for
both use cases
<Luc> +1
Dave: Sent the list another big
question on equation alignment on = signs
... even though unrelated html structure
<liam> +1 alignment
Dave: one of the big, fundamental issues when laying out pages
Markus: Need to decide what to do
next week
... where do you see yourself going in terms of grids?
Dave: Would like to get
illustrations, examples in here. See where they go.
... can start writing up more on hyphenation. WOuld like to
start collecting egs and screenshots
... maybe go into a phase where actively working on a few parts
with lots of input from people with what trying to achieve
Markus: Alan, does that work for you?
Alan: Certainly
Markus: Great. For next week, we need another task force, unless you disagree Dave?
Dave: I need a break :)
Markus: Tzviya, shall we move to your group?
Tzviya: Don't have a lot pulled together, but could review what we have now
Markus: Still a bit fuzzy where
the edges are, need to clarify where to go.
... AOB?
Dave: Several of us will be at digital book world this week
This is scribe.perl Revision: 1.138 of Date: 2013-04-25 13:59:11 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/???/fjh/ Succeeded: s/don't work/only work/ Found Scribe: azaroth Inferring ScribeNick: azaroth WARNING: No "Topic:" lines found. WARNING: No "Present: ... " found! Possibly Present: AWK Alan Ben Bert Bill Bill_Kasdorf Dave IPcaller Ivan Liam Liza Luc Madi Markus P37 P48 Sharad Sharad_Garg Stearns Suzanne TomDN Tzviuya Ugent astearns azaroth azaroth_ benjaminsko dauwhe dictionaries directly dpub dshkolnik etc fjh gcapiel gcapiel1 google joined laudrain lizadaly mgylling plinss problem search tm tmichel trackbot tzviya with You can indicate people for the Present list like this: <dbooth> Present: dbooth jonathan mary <dbooth> Present+ amy WARNING: Replacing previous Regrets list. (Old list: Vlad, Laura, Brady) Use 'Regrets+ ... ' if you meant to add people without replacing the list, such as: <dbooth> Regrets+ Brady_Duga, Vladimir_Levantovsky, Laura_Fowler, Jean_Kaplansky Regrets: Brady_Duga Vladimir_Levantovsky Laura_Fowler Jean_Kaplansky WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Found Date: 13 Jan 2014 Guessing minutes URL: http://www.w3.org/2014/01/13-dpub-minutes.html People with action items: WARNING: No "Topic: ..." lines found! Resulting HTML may have an empty (invalid) <ol>...</ol>. Explanation: "Topic: ..." lines are used to indicate the start of new discussion topics or agenda items, such as: <dbooth> Topic: Review of Amy's report[End of scribe.perl diagnostic output]