Accessibility Minutes 2010 11 03

From MemberWiki

Jump to: navigation, search


  • Jon Gunderson (University of Illinois - Co-Chair)
  • Rich Schwerdtfeger (IBM - Co-Chair)
  • David Todd (IBM)
  • Prasanna Bale (University of Illinois)
  • Ann Abbot (IBM)
  • Marc Johlic (IBM)



Scribe: Marc Johlic

JG: Working on getting test suite examples up

JG: Two main agenda items: Test suite syntax and adding ARIA related rules to ruleset

JG: Will be copying examples over from dev to public site this afternoon

JG: Rule reference = "which IDs will pass this rule / which IDs will fail"

JG: Author of test would put IDs in and then indicate which IDs pass or fail a particular rule

JG: Other type of reference to a test is Calculations

JG: Calculations = "for this ID there is a label and this is what the label should be calculated to"

JG: View HTML source code of page and you'll notice that this information is added as JavaScript data. Rule engines can use this data to validate a rule.

JG: Important takeaway - each test is bound to a rule in db - you can add these rule references and calculations

JG: May be more than OAA wants to do - but could be what W3C wants to do

Rich: Helpful for W3C, but not sure if it's helpful for OAA - might want to bundle it separately

Rich: (unrelated) - Did Nathan get XBrowser stuff in

JG: Nathan created 1 XBrowser rule - alt text for images

RS: Xpath object or what is it?

DT: Thinks they're just extracting out Xbrowser methods

JG: Next step is to talk about process of creating ARIA rules

DT: we have a mix of numeric an camelCase IDs - are we going to one over the other?

JG: Initially rules will be sequentially numbered

DT: Documented on website that particular namespaces for particular vendors

JG: We need to discuss process of creating rulesets

JG: For each ruleset or rule map - Each rule would have a severity code, priority code, & message code

JG: message code would be an ID that can be used to get a message screen in a language

JG: Priority would be P1 P2 P3

JG: For severity code we have violation, potential violation, recommendation,

JG: We should give some level of guidance to developers on how to interpret combinations of severity codes and priority codes

DT: Would seem that violations should come before recommendations

JG: So maybe Recommendations P1, P2, P3 and then Violation P1, P2 P3 etc

JG: Also should determine which Violations take precedence over each other

RS: Customer may want to adjust priority

JG: So maybe more relative - so in a particular criterion some rules may be more important to implement than others

PB: If priorities are left to customer decision, do we need to set priority levels

JG: That option may be for the more advanced customer - suspect that 60% - 80% of the people would use the priorities we set

JG: Question on message codes. Currently they point to dictionary to a dictionary - any idea why we created this second level? Will there be other things that a rule will want other than the message?

search for "rules :"

search for "message-1"

Dictionary key into another dictionary - and only thing in there is 'message'. Does anyone know what other things we'd have in this dictionary other than 'message' - seems like extra indirection.

No one sure where it's derived from - may need to hear from Nathan

JG: May be extra calculation if we don't need it, but may be just for consistency with requirements list

JG: Any objections to check this back in?

No objections on call

JG: From IBM's perspective, do you think people will want to hand code rulesets and use SVN to share them? Likelyhood of using web based tool?

RS: Having web based tool to test it out is a good thing, but want to have SVN to check it in

JG: How do we knew when we have the Open Ajax ruleset 1.0? When do we say this is something OAA publishes?

RS: We need at least one language

RS: For things we can't do manually, we have to put a message out for things we can't test - and those will be part of rules

JG: If a page does not have a feature (i.e. images) do they pass the rule or is it N/A

RS: Would say that's a pass

AA: We currently report that it's N/A - believe the tool should report N/A and not just PASS

JG: Should be easy to say N/A, but rules themselves would never run because there is no node to run them on

JG: How much of ARIA rules do we want in 1.0 - most rules now are HTML

JG: For example currently we don't have a rule that says "every page should have a 'main'

RS: IBM is requiring that pages use a 'main'

and is a WCAG rule - so would be a Violation

JG: So we need a rule

RS: Since we're supporting WCAG 2.0, we should have 'main' content and it should be in our rules

DT: If there are two ways to do it - do we want to push the ARIA technique?

RS: If we're talking about HTML 4, then I prefer ARIA technique

RS: If you have a native host language feature that does the same thing - that's just as good

JG: So for version 1.0 we want at least a manual check for every WCAG requirement?

RS: Need some sort of manual notification for things that can only be done manually

JG: Primarily focus on HTML 4 rules with some ARIA

RS: We need to make sure (for ARIA) that the structural stuff is in there and for proper ARIA attribute usage

RS: We should support IE9 and FF

Next week's meeting will be on Thurs - then we will return to the normal Wed time

Personal tools