Precisely what you're trying to say we need to do: allow for navigation
other than row + 1 and col + 1. If we have a table which has a cell at
r1c2 but no cell at r2c2 and the user navigates down by row from r1c2,
how does an AT find the next appropriate cell? There are two choices
right now:
1. Keep trying IAccessibleTable2::cellAt with increasing row numbers
until we find a cell. But what if the next cell is at r1000c2 (which is
entirely possible with these new ARIA properties)? That'd mean calling
cellAt 999 times.
2. Use accNavigate to walk through the rows, query the first column in
each row to get its row number, then use cellAt to get the cell n that
row in the original column. This does allow us to skip rows which have
no data at all. However, it's still horrible because:
* There might be interspersed thead/tbody/rowgroup elements, requiring
hierarchical tree traversal.
* There are still a huge number of calls.
Aside from the complexity and performance horror, it's also difficult
for implementers to grasp. They look at IA2 and see a table interface,
but then have to understand that actually, the table interface isn't
enough to navigate tables by row and column; they might actually have to
do some insane tree walking to get anything useful.
Things get even worse if you believe an AT should do something other
than skipping rows if there is no cell in the same column on the next
non-empty row. Dominic suggests we should just go to the first column. I
think that is a bad user experience and breaks a user's spatial feel of
the table. So if we want to move to the next "nearest" cell, now we
can't even use cellAt any more. We have to walk the children of every
subsequent row looking for "nearby" coordinates.
Jamie
On 6/01/2016 11:39 AM, Richard Schwerdtfeger wrote:
What are the use cases?
Rich Schwerdtfeger
Inactive hide details for James Teh ---01/05/2016 07:15:27 PM---My
point is that we don't currently have a way of handling thisJames Teh
---01/05/2016 07:15:27 PM---My point is that we don't currently have a
way of handling this decently with IA2 and that we need
From: James Teh <[email protected]>
To: Richard Schwerdtfeger/Austin/IBM@IBMUS
Cc: IAccessible2 mailing list
<[email protected]>,
[email protected], Dominic Mazzoni
<[email protected]>
Date: 01/05/2016 07:15 PM
Subject: Re: [Accessibility-ia2] aria-colcount and aria-rowcount
mapping, again
------------------------------------------------------------------------
My point is that we don't currently have a way of handling this
decently with IA2 and that we need new API calls. Saying "just faf
around with accNavigate and try all sorts of different stuff until you
find what you want" isn't good enough IMO, not least because it's
horribly unperformant cross-proc. If we're going to have gaps at the
start/in the middle of rows, we need some sort of table navigation
function that can handle different types of navigation. Currently,
there is no such function.
Jamie
On 6/01/2016 11:00 AM, Richard Schwerdtfeger wrote:
Jamie,
I have to agree with Dominic that you have a lot of different what
ifs for determining where to go when a cell is empty. I think you
need to be able to pass a parameter that asks the browser to go
where you want based on a set of criteria.
For example, So, what if several rows are skipped? When does the
situation become problematic?
What condition do you want to be met?
e.g. a parameter that tells us what to do if an empty cell is
encountered:
GOTO_FIRST_NONEMPTY_ADJACENT_CELL
What I am getting at is what is the algorithm you want the user
agent to follow when an empty cell is encountered. Does it Vary
based on a parameter. Do you want hidden cells to be ignored?
Rich
Rich Schwerdtfeger
Inactive hide details for James Teh ---01/05/2016 04:50:38 PM---On
5/01/2016 7:53 AM, Richard Schwerdtfeger wrote: >James Teh
---01/05/2016 04:50:38 PM---On 5/01/2016 7:53 AM, Richard
Schwerdtfeger wrote: >
From: James Teh _<[email protected]>_ <mailto:[email protected]>
To: Richard Schwerdtfeger/Austin/IBM@IBMUS, Dominic Mazzoni
_<[email protected]>_ <mailto:[email protected]>
Cc: IAccessible2 mailing list
_<[email protected]>_
<mailto:[email protected]>,
[email protected]_
<mailto:[email protected]>
Date: 01/05/2016 04:50 PM
Subject: Re: [Accessibility-ia2] aria-colcount and aria-rowcount
mapping, again
------------------------------------------------------------------------
On 5/01/2016 7:53 AM, Richard Schwerdtfeger wrote:
Although it is possible to have 2 cells I don't think this
even close to the norm.
Two cells was just an extreme example to demonstrate the issue.
The point is that gaps at the start or middle of a row are
problematic.
Why would we we not just have the screen reader ask to go to a
row and column number?
Because if there is a gap, incrementing row or column by one isn't
sufficient. It could be 1, 2 or an arbitrary number.
It is not difficult to ask the user agent to navigate to
compute the effective cell to the right and ask for the user
agent to navigate there.
It's probably always true that navigating to the adjacent cell in
the next column is as simple as accNavigate with NAVDIR_NEXT.
However, navigating to the adjacent cell in the next row is not so
simple. What if several rows are skipped?
We can add a parameter that states to navigate to the first
non-empty cell to the ... or whatever algorithm we choose.
Add a parameter to what? I think this would require a new
function, which is what I was suggesting.
Dominic wrote:
Jamie, as an alternative, couldn't the screen reader try
navigating to the next / previous cell using the row, column
index, but if that fails, fall back on DOM navigation?
For example, given a current cell (x, y), if the user wants to
go right - to the next cell in the same row, try (x + 1, y)
but if that's empty, use the accessibility tree to navigate to
the next sibling. I suppose we could do that, but it's pretty
inelegant. Oh look, we
have a table interface... but uh, sometimes it's not very useful,
so you might need to use tree navigation. It's not so bad for
columns i guess, but see below for rows.
Similarly, given (x, y) if the user wants to go down - to the
same column in the next row, try (x, y + 1), but if that's
empty, use the accessibility tree to navigate to the next row,
Navigating to the next row has two problems:
1. It's not a simple navigation, since there might be
thead/tbody/rowgroup elements to think about. Sure, it's just a
depth-first traversal, but now we're starting to get fairly
complicated for what should be a simple operation.
2. What if the next row doesn't include that column, but there is
a row 3 rows beyond that that does? Surely we should move to the
row with the matching column. That's not possible with your
algorithm.
then try column x in that row, and if that fails, navigate to
the first cell in that row. We never want to fall back to the first
column like that. If a
user moves down a row, they expect to move down a row, not down
and across (excepting column spans).
Jamie
--
James Teh
Executive Director, NV Access Limited
Ph +61 7 3149 3306_
__www.nvaccess.org_ <http://www.nvaccess.org/>
Facebook: _http://www.facebook.com/NVAccess_
Twitter: @NVAccess
SIP: [email protected]_ <mailto:[email protected]>
--
James Teh
Executive Director, NV Access Limited
Ph +61 7 3149 3306
_www.nvaccess.org_ <http://www.nvaccess.org/>
Facebook: _http://www.facebook.com/NVAccess_
Twitter: @NVAccess
SIP: [email protected]_ <mailto:[email protected]>
--
James Teh
Executive Director, NV Access Limited
Ph +61 7 3149 3306
www.nvaccess.org
Facebook: http://www.facebook.com/NVAccess
Twitter: @NVAccess
SIP: [email protected]
_______________________________________________
Accessibility-ia2 mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/accessibility-ia2
