Accessibility Minutes 2009 09 09
From MemberWiki
Contents |
Participants
- Jon Gunderson (University of Illinois)
- Nathan Jacobiak (ParaSoft)
- Michael Squillace (IBM)
- Vio Onut (IBM)
- Rich Schwerdtfeger (IBM - chair)
Minutes
Validation Rule Codification Requirements
Validation Rule Codification Requirements
Mike: My Post on Validation Rule Codification
Mike: What I wanted to point out is that I sent out 3 different emails that did not get distributed by the OAA server
Mike: They did show up in the wiki as received emails
Mike: The focus in the wiki is based on the last couple weeks discussions
Mike: I want to talk about rule properties and applicability criteria
Mike: I have incorporated Vio's tagging notion in terms of the context
Mike: There is a piece on the rule logic
Mike: The first thing to do is to look at the wiki and modify, change, etc. the wiki.
Jon: the main thing about the a11y extension. If you are interested in headings and need to get the headings for the page and you run individual heading rules, how do you see the data collection fitting into this.
Mike: I have four possible contexts. What Sandy has I think is rule groups which are executed in the context of particular elements of the same kind
Jon: We create a data structure of the header information. then we run our header rules.
Jon: The code that runs the actual logic that the rule runs agains
Mike: The context would be a certain data collection
Jon: the rule context is right. There would be a document level rule but an intermediate process is required.
Jon: The rule logic needs a pointer to a data structure
Mike: The way the rule logic is executed on a particular rule. There is no reason this could not apply to a collection of elements
Nathan: It appears that you guys are talking about the same thing
Nathan: It is identifying all rules of a particular type and then it is running the test against that set.
Mike: what jon is suggesting is collecting all the rules first or are you collecting the roles and elements first and running against the rules.
Jon: we would need to probably update the data structure first.
Jon: perhaps this is a subset of the rule logic.
Jon: Some rules, such as alt tex missing on an img element, you don't need to collect a lot of information
Mike: ... but you could
Mike: I see your point. If you look at our rules you just take individual elements. each element maps to a rule logic ...
Mike: What you are suggesting is that an engine could pass a data structure representing a list of HTML elements
Jon: For an individual rule if the data structure or create one and then run the rules
Jon: perhaps it is just an implementation detail.
Mike: we would need to distinguish between new and old collections.
Jon: create the data structures and run the rules against them
Mike: it gets more complicated if writing against a data structure logic as the rule is expressed in terms of the data structure
Nathan: I think we need to keep these things separate. It is generally an implementation detail. If you write the rules for a given node you can be flexible in your implementation
Nathan: I think Sandy's data structure is different than just a collection of rules
Nathan: It would be better to express the rule logic in therms of a single node and if want to process of collections they can
Mike: Sandy's rule does this.
Mike: there are individual rules that are executed
Rich: for a given event we should probably scope what it being tested
Mike: We talked about a tabular view of what is tested based on the event
Rich: we can scope what is tested on an event in terms of DOM traversal.
Vio: I think I agree with what is said.
Jon: I think there could be a rule group. This could be an option.
Mike: Sandy's notion of a rule are related to a data collection.
Jon: a rule group is optional
Jon: a rule group is one that adds a common data structure
Mike: So we have individual rules and rule sets
Mike: A rule set embodies a specific spec. or guidelines
Mike: A rules set is based on a particular standard
Nathan: For us the rule set is very general but it may or may not be tied to a standard
Mike: The semantic set is a particluar set of guidelines?
Vio: yes
Nathan: yes, but does not have to be.
Nathan: a collection of rules is fine.
Jon: Hopefully buy next week we can show you our web site
Jon: we have different rule sets based on standards and the question is how to repackage
Jon: our approach is we have these sets of rules and how do we map them to the standards
Rich: supposed to float a draft of the 508 refresh at the end of this month
Mike: we want to reuse the across the different guidelines and reports the user wants met
JonG: Mike: What else do we have here
Mike: I am working on rule set information
JonG: Mike: We need to include severity maybe determined by the rule set, one set might be a violation and another maybe a recommendation
Mike: There is a notion of localization, been talking to Dojo team
Mike: They have a server that uses JSON server to get localization information
Mike: Does Firebug have a localization?
Jon: yes
Nathan: for me, having the rules split out into the file makes sense but I don't want them running on the sever for performance reasons.
Nathan: we need to be able to switch things out based on the locale we are working in
Nathan: I want nothing sitting on the server
Mike: the converse is loading all the languages on the client is a hit to load time.
Vio: I second that. Being able to load based on locale is important. Storing things on a server makes sense.
Nathan: Dojo stores locales on the server based on what is needed.
Rich: So loading locales and caching as needed makes sense.
Mike: for our purposes, how bundles are retrieved and how they are represented is an implementation issue.
Mike: there are 2 sets of things that need to be localized: parts of the rule logic and what errors/warnings are reported
Vio: We might have a flag that indicates something is localizeable.
Mike: like an identifier?
Vio: yes
Vio: Some rules may not be provided for a specific language.
Vio: I am referring to the parameters that go into rules
Vio: they would need to display the right language
Vio: some rules may not need to change the logic.
Vio: I see a decoupling of what is displayed to the user and the rule itself.
Vio: The bottom line is that if there is a rule that needs globalization it should be flagged.
Vio: that will help
Mike: the other option is that you would not need a parameter. The parameter format could have a language indicator.
Mike: I am going to work on the wiki this week.
Mike: I want to include our 10am discussions
Jon: I did some rules and rule logic.
Mike: Next week will have the rules discussion on the Wednesday call next week. Mike to chair next Wednesday.
Reporting Best Practices
Accessibility Reporting Best Practices
Rich: the ARIA subteam addressed the problems we found in the containment model
Rich: Rich will update the JSON file on the ARIA pattern matching
Rich: to look at processing elements whose meaning is determined by host language attributes aka: input type="" ...
ACTION ITEM: Group review Rule Codification requirements for next week