- From: Andi Snow-Weaver <andisnow@us.ibm.com>
- Date: Tue, 13 Aug 2002 18:37:46 -0400
- To: w3c-wai-gl@w3.org
Present: jason, gregg, andi (scribe), ben, bengt, loretta, john, avi, eugenia, lee, preety kumar Regrets: Wendy, Gian, Matt, Paul, Doyle jason - success criteria for checkpoint 5.2. Need stronger backward compatibility requirement at level two. lee - Prefers to limit backward compatibility to one level previous version. Lot of engines for AT are IE. Some are starting to swing over to Gecko format. Drive people to upgrading their software for the free stuff. JAWS will use latest version of IE that it finds on the system. jason - level 1 specify baseline, level 2 should be more stringent, level 3 goes beyond that. Open as to what these should actually be. john - main difference in two proposals is that one is keyed to timeframe something has been available and other is keyed to version level of software. jason - sets minimum technological threshold. Content must still work below the baseline but may not work optimally. Has impact on what standards and technology people rely on. Need objectives for what we are trying to achieve at level 2 then work out criteria to achieve those objectives. lee - FLASH can be accessible if use Window Eyes. What Windows platform does this require? Windows 2000 or XP? or Windows 95/98? john - most developers use version numbers, not timeframes gregg - vhs won over Beta because there was more of it. Worry that we are defining compatibility in terms of what there are the most of. lee - primary drivers are IE and Gecko. If we design specifically for IE, there may be problems down the road. JAWS, WE, and HPR all use IE. If also works for Gecko, covers most of the rest. gregg - some features exist in one but not the other. If the criteria is it must work in both, these types of features would be excluded. jason - could clarify that there have to be "interoperable" features. gregg - so longdesc would be illegal john - HPR and JAWS support it. gregg - HPR and JAWS don't count because they are both on same engine lee - Gecko supports longdesc lorreta - it will support it when it comes out. but requirement is that it be supported in the previous version too. loretta, andi, and john agree that it is easier to attract versions than timeframe lee - can we say "engines" instead of "versions" to avoid possibility that two options chosen are JAWS and IE because JAWS uses IE engine jason - proposal worded as "component" because it depends on where the feature is implemented gregg - need to determine if can't figure out how they would do it. lee - benefit is to get people to stop supporting just IE in their designs john - what does below the baseline mean? jason - level 1 criteria talks about technologies "above the baseline". These are technologies you may use but are not relying on. Level 2 talks about the technologies you are relying on. andi - below the baseline is what is "required" to use the site gregg - level 1 page is still usable if technologes that are not in the required list are not available. level 2 says that the technologies you choose for the required list have to have been available in at least two implementations for at least two versions. gregg - how do you get two independent implementations of FLASH or other proprietary implementations? gregg - what is the reason for having two implementations? lee - could provide note on third party proprietary technology gregg - that's going backwards lee - want two user agent engines loretta - assume the reason we would ask for two implementations is to ensure that specification is adequate. PDF has public spec. GO Script supports it but it is not accessible. jason - requiring that technologies you are relying on are public and well supported. Combination of interoperability and backward compatibility help ensure that you have a wider variety of users and devices that will be able to access it. Gregg's question is what is the rationale for that type of requirement. jason - AT support is covered in another checkpoint. loretta - what was goal of this checkpoint jason - trying to avoid lock-in to some particular version of software gregg - what is it that we are trying to address with this checkpoint? Level 1 is just letting people know what you require. Need to keep the word "accessibly" in there. jason - worried about using the word "accessibly". These guidelines define what "accessible" means. 5.3 covers the AT problem gregg - then what is the purpose of 5.2? Backward compatibility. What does two independent versions have to do with backward compatibility? What does defining your minimum requirements have to do with backward compatibility john - level 1 does talk about backward compatibility gregg - title must be trying to get at something different. john - level 1 is to have a baseline and state what it is. level 2 is that the baseline has to have been available for at least two versions jason - don't want to require any level of backward compatibility at level 1 but ask that people be concious of what they are requiring. jason - level 2 is to place stronger restraints on the required list gregg - proper title might be "technologies required in user agents in order for content to conform with guidelines are declared and commonly available" gregg - this is the first checkpoint where we have required metadata or policy statement on the site jason - before we put it on level 2 because checkpoints were judgement based gregg - will post re-write of success criteria pk - what is meant by "versions"? are these major releases like 4.x to 5.x or would 5.4 and 5.3 count as two "versions" gregg - how will page authors know how long features have been available or which features are in which one? gregg - how important is this? If we don't have this in here, will pages not be accessible? lee - want to get users to upgrade their free stuff. john - costs money if they have to upgrade their hardware jason - ultimately implementors will make a decision about how far back they want to go. john - we have responsibility to help people figure this out jason - even if we use time period instead of version, somebody still has to provide the information. lee - don't know anyone who keeps track of technologies based on time frame but there are people who keep track of it based on versions. jason - are we any closer to resolution? gregg - think we have a better title. Need to post notes to mailing list. gregg - need to examine what we are trying to achieve and ensure that success criteria that achieve that and are not treating a symptom. lee - think we have done it when we say "two implementations". will open it up to force developers to design to more than just IE gregg - questions about second topic about how we pick success criteria gregg - decided that success criteria have to be things that are testable. can we further state that this is machine testable or HIRR - high inter rater reliability; i.e. most raters who understand the issue and the measure would all say the same things avi - this gets tough with 4.1. trying to amass some things that would be testable. gregg - level 1 and level 2 would have to be applicable to essentially all content on most all types of sites, all level 1 criteria have to be met before anything in level 2 can be claimed. [group consenses] gregg - level 3 is for people who are pushing ahead and does not have to apply to all types of sites jason - if we have a lot of these that don't apply to most sites, we should question whether or not we should have them john - isn't it possible to claim a particular level of conformance for a particular page on a site gregg - yes, but that's not how they usually get applied avi - technical standards fit this but may not apply for content type checkpoints (4.1 et al) [gregg will post modifications] avi - one of challenges is how to write 4.1 in a way that is reasonable and pragmatic. May not be reasonable to have to make all of your text comply. Andi andisnow@us.ibm.com IBM Accessibility Center http://www.ibm.com/able http://w3.austin.ibm.com/~snsinfo
Received on Tuesday, 13 August 2002 18:35:03 UTC