2006-130 Meeting Minutes

PDF-UA 2006-129
September 20, 2006

PDF/UA Working Group Teleconference

PDF/UA (Universal Accessibility) Working Group Draft Meeting Agenda
Wednesday, September 20, 2006, 3:00 P.M. - 5:00 P.M. EST
Conference Call Information
International Dial-in: +1-303-928-2600
U.S. Dial-in: 1-888-742-8686
Conference ID: 7083945

NOTE: The meeting will be conducted "live" on the Wiki.

Roll Call

Duff Johnson, DJ - Chair, Document Solutions
Joe Clark, JC - Independent Accessibility Consultant
Mark Gavin, MG - Appligent
Loretta Guarino-Reid, LGR - Adobe Systems
Dick Herring, DH - Independent Accessibility Consultant
Greg Pisocky, GP - Adobe Systems
Neil Soiffer, NS - Design Science

Approval of PDF/UA 2006-129 Agenda

Guarino-Reid / Soiffer

Approval of 2006-128 Meeting Minutes

Johnson / Soiffer

Report on / update and discuss Action Items

The Action Items were updated appropriately.

Greg Pisocky to update the multi-media section.

Abbreviations and Acronyms to be updated to show which subject will be discussed at the next meeting or we choose to do it now. Natural Language specification leaf to be topic for next meeting. Discuss Neal Soiffer's issue, assurance that trip be productive. List of issues needs to be resolved.

DJ: Indicate ideas for Toronto, place a comment into the agenda page. Perhaps address concerns regarding MathML. Agenda will be on the meetings page.

JC: For the face to face, is it possible to have on hand printed copies of the PDF Reference?

GP: I can see that those are on hand.

JC: 1st item will be math.

DJ: Best way to do this is to present the page and invite people to comment on.

NS: Let's identify the finished topics.

DJ: Other suggestions for face to face should be added as a comment to Face to Face Agenda page.

Report on status for "Assigned Leaves"

Five minutes on Logical Structure Page.

NS: Only leaves of trees point to content

LGR: That is true

NS: Can the formula tag and the math tag

DJ: Suppose everything is in a Math tag, where the heirarchy of the tags has operational significance for the screen reader

LGR: I don't see there is any problem here, as that is true with tables.. Presumably have to make some decisions is formual tag child of a math tag and vice versa.

NS: So for example, suppose I have headings for Sections... How does it look?

LGR: I think of a Section as being a container. It's not a leaf, but if you keep burrowing down in that heirarchy, sooner or later there is content. Headings are an awkward example, tagged PDF adopts the HTML model where it is floating. We can envision a world like XML or DAISY.

NS: In the current PDF model headings are siblings of paragraphs

LGR: The current PDF model does not speak one way or the other about that. This gets us into the discussion of looking at the current definition of tagged PDF. The problem is, for lots of documents that you would like to propose there are differences that do not fit that schemea.

NS: Which would be tghe purpose of a standard, to tighten things up a bit.

LGR: I expect to go looking into the tags one of these days.

DJ: There's the opening. Set up a container and set up with a piece of content, have the AT see the container and change state when it's done.

DJ: Interesting question, how or whether to organize PDF tags.

NS: The concern is having 2 tags (say a formula tag) simple math and formula are just containers which encompass larger items.

LGR: I actually think that's okay. No reason you can have one strucutre tag which contains one tag which is another structure.

DJ: We can develop a convention where there is a wrapper tag for math <Formula> for example.

JC: Could NS produce a wriiten plain text example of what we are discussing.

James Cole: Document Outline (Bookmarks)

Angy Brooks: Page Labels, Articles, Presentations

Greg Pisocky: XFA Forms

Richard Herring: Media Equivalents

Melanie Zwack: Tables

Ferass Elrayes: Content/Structure Separation

(Those without current "leaf" assignments - pick one, or be assigned one!

Discuss Logical Structure

Loretta has provided a section reference.

JC: Order in which things are read

LGR: Logical structure chapter describes the mechanism inside trhe PDF file that holds the logical strucutre ifnroamtion. It is a data structure. One we happen to use for a particualuar purpose. You can infer a reading order from it, but I am not sure

NS: One example, if you have a parent, what is the nature of it's reading ordcer to the children

LGR: So now yhou are taliking about a client. Ther is not one single answer. Suppose you have an article that goes on for several pages. You might have an article tag which has several children in the order that is implied there.

JC: Because we have this concept of logical structure as oppoesed to linear order w can solve problems So we can ....

LGR: 2 edged sword, we have the flexibility to present the logical structure that is differnect than the physical presentation Yhre Logical data styructure, that mechanism

DJ A concrete example. Example one word prior tothe next

LGR: we do not take logical astructure to the level of word by word.

DJ LS = define a paragraph Tagging structure = sequence

LGR Logical structure is a tree, at the leaves are content on the page and the kinds of content that can be attached is defined in the reference. THe reading order is deriveed for any given logical structure you can walk the tree and visit the leaves in the order you find in the tree. That is what your AT woudl read if it does not know any bettter. YOu m

NS: For exsazmple tables. Some structurel but rows and colsumns could be listed in an order

LGR: Someone may want to see the columns first, buyt the tagging specifies row order. One of the things that is important for us which defines a set of elements, a set of attributes, and relationships for which they should be interpreted. Unfortunately it's a confusing relationship.

GP: Best possible

LGR: You might want to say that they are the same or consistent. We would need to look at some cases carefully. To get some of the physical effeets, sometimes it's necessary to render characters in other than the reading order to get the physical effects you are trying to achieve.

LGR: What we really need is the logical structure tree to reflect the logical structure of the data.

Tagged PDF is different from logical structure.

LGR: Logical structure is a general mechanism, Extensibility was important. Neither the set of structure elements nor the set of attributes is predefined. The mechanism is gerneal and extensible.

NS: The standard ones doe have some special status (10.8)

LGR: They have speical status, the label tagged PDF aas used by Adobe referes to that set of tags. Clients at least have that definiiton for that set of tags.

DH: Which special tags?

LGR: THose oare the tagged PDF tags. (actually 10.7 Tagged PDF) We can define a set of tags. In fact I expect what we need are tables that look like the ones in 10.7 I think the trouble in this case is pointing off in PDF it's not clear what these mean. Key, value, type, how is this represented inside the PDF file. How to relate the XML data structure and the PDF data structure. I haven't read the MathML spec so I don't know if there are issues or not.

DH: Is that part of the issue having the freedom to do that

NS: The MathML spec is very long.

LGR: Distinction between XFA and ref MathML. XFA spec We are talking about representing those concepts inside the logical structure data structure.

DH: Standard attribute owners on page 843. RTF, CSS1. CSS2 I am wondering if MathML can be treated the same way. An atttribute objected applied only when exporting PDF content (...go see Richard's reference). We'll pass the stuff along to some other agent, but we cannot guarantee that Acrobat or any other particualar viewer will handle it.

LGR: Go on to next section wehre it talks about attribute valuses and inheritance.

DH: I am looking for a precendent

LGR: I don't know whether it was different or not. I do expect there to be some MathML owner. I don't think this forbids something like a screen reader from using those attributes, We would just have to know the name of theowneer which again we woudl have to put into the spec. THe SR is actually going throught the accessibitliy API. THe DOM API exposes this in full detail. SOmething like MSAA is limited, so there is no way to expose all of that to the API.

LGR: We could define our own sets of tags and attributes that are not part of the PDF spec. Problems with that. We have gone down this path. The easiest thing would have been that the PDF spec was sufficiently broad. THe knothole that things have to be added, we need to extend the current spec. Or take advantage of the inherent extensibility.

NS: Seems to me one of the groups that would review the PDF/UA spec would be the people at Adobe who say what the PDF Standard is.

LGR: I suspect that's why we need tables that will look kind of what you see in 10.7

GP: Would production of these tables be something worth working towards as a deliverable for the face to face.

NS: Seems to me the generic issue of how we're handling math should be settled before we do the detail work.

LGR: Tables, only one half dozen structure elements or half dozen attributes needed to tag a table. THough we never readhed resolution on talbles.

NS: That maybe another item for a face to face.

LGR: About table elements in section 10.7 there is a table element section 10.2.4 and a section on table attributes. THis could all be fit on 2 pages, though I suspect that won't be true for mathML.

NS: Presentation 30, Content 100 or so of those

LGR: Sounds like a strong argument for going for presentation.

NS: A table similar to what's done for the tables would probably be about 4 to 5 pages if we wrote tersly

LGR: I would hope we would write tersely and include a pointer to the MsthML spec.

NS: One suggestion. Perhaps one of the items to discuss next week is to review tables, and prepare something to present. Discussion next week so its fresh on our minds and Loretta and I have some.

AI: LGR to review the MathML discussion. If comfortable, then work through all the low level details. That and tables. I would like us to go through the current tagged definition of TAGGED PDF.

AI: All review the work on tables.

GP: So looking at setting 3 objectives. Work systematically through the definition of tagged PDF, ponder extension of the Tables section, and hammer the math section out.

JC: Anne van Kestrin Opera W3C in Morocco. Willing to look at our tags.

LGR: Procedural question, does he have to join the group?

AI: JC to remind Duff to invite Anne Van Kestrin and see if he can make it to Toronto for Face to Face.

AI: GP doing tables we must invite Steve Ferg at least by speakerphone.

AI: Proposal from DH rather than specify under what circumstances captioning dubbing, should go into informative portions or guidelines, but not in the spec. Rather than write rules that say what you shuld use where, id capabilities that can be used.

LGR: The PDF SPec remains the PDF SPec. THe PDF/UA spec will say for example "you MUST supply the following" from the existing SPec.

NS: Some of the spec is written specifically to talk about what Acrobat can do (Appendix H)

LGR: THings that talk about Acrobat behavior are not something we want to talk about in this spec.

LGR: Critical thing is to get some proposals put forth, I would encourage you to go ahead and put forth as a proposal and suggest it to Duff as an agenda items for one of these meetings.

DH: I like the idea of tracking our progress, as a metric.

AI: DH to propose future topics for upcoming meetings.

Wrapup

Settled upon 3 working items for the face to face

Definition of tagged PDF, identify those items of agreement and distinguish what will require further discussion

Examine current PDF Table tags determine if set is sufficient. Propose

Develop table for MathML presentation tags based on treatment of PDF tables

Next meeting, October 4, 2006 3:00 pm to 5:00 pm Eastern time.

Is a car necessary for Toronto? Won't need it.

DH: PDF can already contain and select audio descriptions, over dubs, captions, subtitles, etc..... I am interpreting the rendition capability as powerful jsut as the spec if written

LGR: True, but unfortunate limitations with respect to implementaiton.

Adjourn

4:58 pm