W3C home > Mailing lists > Public > w3c-wai-ua@w3.org > October to December 2000

Re: Belated last call comments from HTML WG

From: Ian Jacobs <ij@w3.org>
Date: Mon, 27 Nov 2000 16:03:16 -0500
Message-ID: <3A22CC14.58193171@w3.org>
To: w3c-html-wg@w3.org
CC: w3c-wai-ua@w3.org
Dear HTML Working Group,

Thank you for sending in comments. I will incorporate
your editorial comments. 

 >8.4 "This checkpoint does not require that the outline view be
navigable,
 >but this is recommended; " Is this a 'should'/priority 2? Does
following or
 >not following this recommendation affect the A/AA/AAA rating?

This comment is in the Note following the checkpoint, and therefore
is not part of the requirement. Section 2 [1] reads:

   "Informative notes about the checkpoint. 
    These notes include examples, cross references, and commentary to
help
    readers understand the scope of the checkpoint."

I will state more clearly at this point that the notes do not include
requirements that are part of conformance, such as :

    "Informative notes about the checkpoint. These notes do not
    include requirements that must be satisfied as part of
    conformance; they are informative only. They are meant to clarify
    the scope of the checkpoint through further description, examples,
    cross references, and commentary."

[1] http://www.w3.org/TR/2000/WD-UAAG10-20001023/#Guidelines

 >--
 >- if you are going to normatively reference RFC 2119, the words
 >"should", "may", and "must" MUST be uppercase

RFC 2119  [2] says:
 
  "These words are often capitalized."

[2] http://www.ietf.org/rfc/rfc2119.txt

 >- 3.5. instead of "scripts or applets", how about "scripts, or other
 >executable content"?  applets and scripts are not the only form of
 >executable
 >content.

At our recent face-to-face meeting, we decided for issue 364 [3]
to talk about "scripts, applets, and plug-ins". 

PROPOSED: Add "and other executable content" to 3.5.

[3] http://server.rehab.uiuc.edu/ua-issues/issues-linear-lc2.html#364

 >General impression:
 >This document is in general too vague in the requirements for the
different
 >user agents (browser, plug-ins, accessibility application, content
 >provider). In addition, there is no clear indication as which  "user
agent"
 >is responsibility for satisfying each requirement. In reading sections
1 &
 >2, it is extremely confusing as to which "agent" is responsible for
what
 >action/behavior. 

Conformance (which is essentially about responsibilities) is discussed
in section 3. Please indicate whether after reading section 3, you
feel that it addresses your issue sufficiently.

[snip]

 >Other items that came to mind just in the course of reading the
document:
 >
 >What disabilities? Are they talking about physical and mental?

The document proposes cross-disability requirements (visual, hearing,
physical, and cognitive). We should state that up front (e.g., in the
section on "scope"). For instance:

   "This document was designed specifically to improve the
   accessibility of mainstream user agents with multimedia
   capabilities for users with one or more disabilities (visual,
   hearing, physical, and cognitive)."

 >The document continually referred to prefs that should be set, but
state in
 >the beginning that the application should work on default settings --
but
 >what disability should the default be defaulted to? Which doesn't seem
 >correct does it? So, the default install is what then?

I will add this question to the issues list as issue 447:
  http://server.rehab.uiuc.edu/ua-issues/issues-linear-lc2.html#447

 >How are files such as PDF to be handled? The data is read-only, so how
could
 >that be interpreted?

I believe that issue 324 [4] encompasses this issue, and this was
discussed at the WG's recent face-to-face meeting [5]. I don't think
that the "read-only" issue affects the UAAG 1.0.

[4] http://server.rehab.uiuc.edu/ua-issues/issues-linear-lc2.html#324
[5] http://www.w3.org/WAI/UA/2000/11/minutes-20001116#issue-324

 >More specific input:
 >
 >section1.1
 >first list
 >   list item (1)
 >       what does "process some types" mean, this is unclear

Intended to refer to cognitive disabilities. I don't think
"understand" is the appropriate term. 

 >   list item (2)
 >       this is clearly beyond any user agent and is more the
responsibility
 >of the content provider

Point well-taken. This section currenly lists general user needs, not
all of which are meant to be satisfied by user agent. However, rather
than eliminate the need in question, I'd prefer to clarify that some
needs will be met by a conforming user agent, while others should
be met by authors or specification writers.

 >second list
 >   list item (1)
 >
 >pre-processing knowledge is beyond the scope of the browser,
interpretation
 >of internet connectivity, bandwidth, etc. is unrealistic

That depends on the format. For instance, SMIL 1.0 and SMIL 2.0
include parameters that allow the author to specify different
content as a function of available bandwidth (and user preferences).

 >section 1.2, 4th para, note

Text in question:

  <BLOCKQUOTE>
  Note: User agent developers are strongly encouraged to
  design software that conforms in the default configuration. Users
  may not be able to install complementary software because the
  default configuration does not allow it easily (e.g., the mechanisms
  for retrieving and installing plug-ins are not accessible by
  default), because they don't have access privileges on a public
  computer, etc.
  </BLOCKQUOTE>

 >
 >forcing the usage of a particular plug-in, or secondary application
should
 >not be the role of the browser.

That's not a requirement of this document. Can you explain from whence
you draw that conclusion?

 > Users have specific likes and dislikes in
 >regards to software applications. Platforms also play a major role in
 >determining what secondary applications can be used. It would seem
more
 >appropriate that the request be that the user agent provide a
mechanism in
 >which the user may easily utilize applications of choice and that the
user
 >agents provide a "seamless" integration.

This note is primarily about being able to install the
software. 

I'm not sure how a requirement to "easily utilize applications of
choice" would be phrased (and observable). Can you provide some
examples?

 >Guideline 1
 >General comments:
 >
 >operating systems must provide a mechanism to accomplish much of this
 >requirement.

Yes.

 >how should applets, object and images be treated

I'm not sure what you mean. Can you explain further?

 >checkpoint 1.3
 >
 >is platform specific operations more important than cross-platform
 >consistency? 

There are advantages to both for assistive technologies, or ATs. This
is discussed in the definition of "API":
 
  - Cross-platform solutions benefit ATs that work on different
platforms.
  - Platform-specific conventions benefit ATs that work with other
software
    than user agents (e.g., text editors, spreadsheets, etc.).

 > It seems that consistency would be of greater importance,
 >especially if the user utilizes multiple platforms, thus increasing
 >confusion to the user
 >
 >Guideline 2
 >para 1
 >
 >how would user agents, i.e. the browser, provide different content
modes?
 >Shouldn't this be provided by the content provider?

The statement isn't that the user agent must provide output in
different modes, only that users require output in different modes.

 >checkpoint 2.2
 >
 >is the expectation that functions such as redirects also be included
in this
 >requirement? What if the timer is server based?

Good question. Please see issue 360 [6]. We discussed this at
our recent face-to-face meeting [7] and the WG decided that
server-side timing control was out of scope for this document and that
that question should be pursued in a later version of the document.

[6] http://server.rehab.uiuc.edu/ua-issues/issues-linear-lc2.html#360
[7] http://www.w3.org/WAI/UA/2000/11/minutes-20001116#issue-360

 >checkpoint 2.5
 >
 >this is too vague and needs further explanation

We will be working on this checkpoint as a result of other
discussions. However, if you wish to suggest how you feel it
should be clarified, we welcome additional detail. Otherwise, I
don't know how to address the comment that it "needs further
explanation".

 >section 2.7

Please see WG discussions of issue 329 at our recent face-to-face
meeting [8].

[8] http://www.w3.org/WAI/UA/2000/11/minutes-20001116#issue-329

 >is BDO also included in this? 

Yes, in the case of HTML. That's also covered by the requirement
to conform to specs (checkpoint 6.2).

 >What about charset selection? 

We are discussing a possible new requirement specifically about
charset support. Refer to issue 327 [9]

[9] http://server.rehab.uiuc.edu/ua-issues/issues-linear-lc2.html#327

 > What about entity usage?

That seems to be an implementation issue: the character is
just represented in another way than directly in bytes. 

PROPOSED: Mention BDO, charsets, and entities in the techniques
for this checkpoint.

 >Guideline 3
 >
 >General comment: interpretation of imagemap coordinates in a visual
manner
 >would be overly taxing for the user agent, the source is known and
could be
 >easily extracted for interpretation, hence displaying a list of
pointers,
 >but not interpreting via the layout. That approach seems to be more
easily
 >adoptable.

I'm not sure I understand your comment. Could you please explain to
what requirement it refers? I don't see any requirements related to
image maps in Guideline 3.

 >In addition, it seems that "preprocessing" of a page will result in
resource
 >constraints and will increase the time to load a page. Is timing not
an
 >issue for rendering information?

I'm not sure to understand your point (are you suggesting a new
requirement?). I think that in general, performance issues that affect
everyone equally are not specifically accessibility issues. In
general, we don't include requirements to solve problems that affect
all users (including users with disabilities).

 >Guideline 4
 >2nd para
 >
 >this seems to be more content provider requirements

That depends on the format. It is helpful for authors to be
aware of this, but since some formats (e.g., SMIL) allow the
specification of synchronization information explicitly, the
user agent can "know" enough to provide additional control
automatically.

 >checkpoint 4.6
 >
 >this is specific to plug-in or accessibility application, where such
 >requirements are listed, it will assist in alleviating confusion if
the
 >specification pointed to a specific "user agent" -- otherwise it is
unclear
 >as to which software should be responsible.

This is a conformance issue. If the conformance claim is "this browser
plus this plug-in (for checkpoint 4.6) meet the requirements of the
document", then there is no problem.

 >section 4.10
 >
 >this seems specific to the OS

That may be the best way to satisfy the requirement.

 >Guideline 5
 >
 >General comments:
 >
 >what is meant by timely? 

This is indeed a hard question, and the Working Group is aware
that it may be difficult to verify conformance to this requirement -
it may vary from system to system, for example. Refer to issue 127 [10],
from the previous last call. The WG decided that this performance
requirement
was important, and that the techniques document would give a sense
about what performance is expected.

[10] http://server.rehab.uiuc.edu/ua-issues/issues-linear.html#127

 > What is timely to me, may be unacceptable to
 >someone else. Be specific, give a time range that is deemed
acceptable.
 >as mentioned above, is platform specific behavior more important than
 >consistency across platforms for the user?
 >much of the information listed as requirements is unclear as to which
 >application or category (browser, plug-in, accessibility application,
 >content provider) is responsible.
 >
 >checkpoints 5.1 & 5.2, Note
 >
 >These specifications are defined the ... -- should be "These
specifications
 >are defined within the ..."

I prefer "by the". However, some edits are required due to the way
in which the DOM specs were split recently.

 >checkpoint 5.7
 >
 >is this section referring to viewing the page or editing the page? 

Good question - is this a read-only or a read-write requirement. I
suspect it is for both, but we have gone to the effort to distinguish
checkpoints 5.1 and 5.2, so I think we should do the same for checkpoint
5.7. Added as issue 448 [11].

[11] http://server.rehab.uiuc.edu/ua-issues/issues-linear-lc2.html#448

> Why would a user need to access the CSS when viewing a document?

This checkpoint is for programmatic access to style information,
which may benefit the assistive technology for a particular rendering.

 >Guideline 7
 >
 >para #1
 >
 >by visual orientation or by structural orientation? 

 Structural. Not everyone uses visual orientation.

 >Visual orientation will
 >be costly in processing the data based on resolution, window size,
etc.
 >
 >para #2
 >
 >how should line numbering be accomplished on wrapped content? The
numbering
 >will be altered based on resolution, window size, etc. Consistently
could
 >not be achieved.

We don't have a requirement for line-numbering, and that seems to
be an implementation issue that is solvable in any case.

 >para #4
 >
 >via what mechanism? Should this be based on element? 

 That's one option, but elements alone may not provide the "proper"
 level of abstraction. For instance, in HTML, it is useful to think of
 headings as containers, not separators, for the construction of
 a table of contents. Thus, in some cases, developers would do better
 applying heuristics rather than sticking strictly to the DTD.

 > What about incorrectly coded data? 

 That's an authoring problem. This document does not make repair
 requirements for this case (refer to "scope", section 1.2).

 > What about body element text? What about context sensitive data?

Some of the questions you ask are left open because developers may
offer different solutions to meeting the requirement, which is spelled
out in checkpoint 7.6 (at least to the best of the WG's ability, after
long deliberations).


 >checkpoint 7.1
 >
 >What about contextual data such as mouse over data? The user
experience
 >could get out of sync with the page. How should this be addressed?

 I'm not sure I understand. First there is an authoring issue: if
 authors provide content that is only accessible through scripts, they
 fail WCAG 1.0 and that's out of scope for UAAG 1.0. Can you explain
 how content that appears at a mouse-over event would constitute
 a new viewport to which the user would have to be able to navigate?

 >section 7.5, Note
 >
 >is this referring to the content in focus, or the page?

The WG will be clarifying that question (the answer is: the whole
page). Refer to issue 354 [12].

[12] http://server.rehab.uiuc.edu/ua-issues/issues-linear-lc2.html#354

 >Guideline 8
 >first list
 >   list item (2)
 >
 >this is content provider issue

But it's also a user agent issue: to make available information
that has been required of the author by WCAG 1.0.

 >checkpoints 8.3, 8.4, 8.5, 8.9
 >
 >checkpoints 8.3, 8.4 and 8.9 are content provider issues, checkpoint
8.5 is
 >partially content provider issues

As mentioned above, this document requires user agents to:

 1) Make available information provided by the author
 2) Calculate some information so that the author doesn't have to 
    provide redundant content. For instance, Amaya
   (http://www.w3.org/Amaya) generates an outline view based on 
   document headings. 

 >checkpoint 9.2
 >
 >each platform has different conventions, resulting in confusion to the
user,
 >when or if they utilize multiple platforms

That is true. I don't believe that resolving conflicts among different
platform conventions is in scope for this document. I think it does
make sense that a user agent running on a particular platform (even if
designed to be able to work cross-platform) should be well-integrated
into that world, which includes observing conventions for that
platform and avoiding conflicts.

 >checkpoint 9.3
 >
 >single key? based on the context of the focus, this could drastically
affect
 >the document, I would suspect that the author is indicating the usage
of the
 >function keys, however, it is unclear.

This checkpoint is not specifically about single-key access, just
about author-specified bindings. Could you clarify your point?

Thank you,

 - Ian

-- 
Ian Jacobs (jacobs@w3.org)   http://www.w3.org/People/Jacobs
Tel:                         +1 831 457-2842
Cell:                        +1 917 450-8783
Received on Monday, 27 November 2000 16:03:17 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 October 2009 06:50:22 GMT