WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: Z-Order and Tag Order Need to Match?

for

From: Christine Hogenkamp
Date: May 28, 2020 1:39PM


Hi all,

I have been reading the replies with great interest, it was only once I got
into making accessible PDFs that I really thought about all the different
programs used to make PDFs and how they might do it differently and how
that affects accessibility.

My little tidbits so far as someone using inDesign and Acrobat to make PDFs:

- There is definitely a lack of streamlined, clear information about making
accessible PDFs in inDesign. Oh there are plenty of pages floating around
on the subject on Adobe sites, but they are in different locations that
come up with different keyword searches, of varying ages and different
instructions, some meant for older versions of CC. It has taken my team a
good number of months to create our own documentation on best practices for
accessible pdfs and even then, there are probably discrepancies between our
steps and whatever we are officially meant to be doing, because our
experiences creating PDFs are unique to us, and also have to match up with
our current in-house design practices. There has been a great deal of
trial and error, and we have a fairly lengthy troubleshooting section to
our doc, to try to fix particular errors that come up in the Acrobat
accessibility checker.

- using NVDA to test PDFs was actually extremely helpful, before using NVDA
we thought that since the Reading Order panel was showing the correct
logical reading order, then the PDF was good to go. It has also helped
demonstrate when elements are not set correctly in less obvious ways or you
would assume would be automatically set correctly. The most recent example
of this was setting alt text on a Table of Contents link to another page.
You would assume turning the existing live text into a link and setting
that link to Page X is sufficient, as setting alt text for text is not
intuitive and seems redundant even though the option is clearly present in
the link options in the InDesign Links panel. But it definitely was
necessary, otherwise the link was read out as "Blank" in NVDA.

- I have made some basic inquiries with different literary assistance
software companies to ask them which elements they use to determine reading
order in a document. Some use the tag tree, some use the Reading Order
panel, and one in particular told me their software took a greyscale
snapshot of the whole page and then used OCR software to convert it into
text (why do they do this when live text already exists was not explained
LOL) so I agree that it would be nearly impossible to create a PDF that was
useable by all assistive tech programs. It would be nice if we could all
agree on the Tag Tree to determine the logical reading order, but the
history of computer related software development is not encouraging on the
"herding cats" front of things. Could we even agree here on this mailing
list that the tag tree is the best method? Ha ha!

- The issue with these automatic checkers, be it Acrobat or PAC, has always
been the tendency for false positives and how difficult it can be to find
solutions for the errors. Acrobat at least tries to give you a link with a
possible solution but I have yet to discover any sort of official resource
for PAC errors to clearly explain what needs to be done, which can make it
difficult to consider PAC a useful tool.

Ultimately, until the original authors of these programs, both Adobe and
whoever made PAC, improve the documentation surrounding their programs, I
would expect PDFs to continue to be made incorrectly. How can the average
user do any better than the program does by default? How could they be
expected to even know let alone understand what is wrong with their PDFs?

Thanks for all the food for thought :)

*===*
*Christine Hogenkamp*
Front-end Developer

CONTEXT CREATIVE
317 ADELAIDE ST. W., #500 | TORONTO, ON CANADA | M5V 1P9
<https://maps.google.com/?q=317+ADELAIDE+ST.+W.,+%23500%C2%A0+%7C%C2%A0+TORONTO,+ON+CANADA%C2%A0+%7C%C2%A0+M5V+1P9&entry=gmail&source=g>

>
>