– DRAFT –
Personalization Task Force Teleconference

<LisaSeemanKest> comments on explainer: https://lists.w3.org/Archives/Public/public-personalization-tf/2021Apr/0023.html

<janina> brb

regrets, CharlesL

Mats discussion on attributes @action, @destination _and_ @purpose. see https://lists.w3.org/Archives/Public/public-personalization-tf/2021Apr/0019.html

<LisaSeemanKest> https://github.com/w3c/personalization-semantics/wiki/Comparison-of-ways-to-use-vocabulary-in-content

LisaSeemanKest: adding some links referencing our ways of using vocabulary

<LisaSeemanKest> https://github.com/w3c/personalization-semantics/wiki/Prototypes-with-data-dash-*-(Take-2)#links-and-buttons--action-destination-or-action--destination

LisaSeemanKest: prev. link has some coding examples of using data-

<LisaSeemanKest> https://github.com/w3c/personalization-semantics/wiki/Use-cases

janina: question came up about whether we considered using computed role from AOM.

<Zakim> janina, you wanted to ask whether we considered using computed role

LisaSeemanKest: are there other issues to bring up

JF: wants to discuss true and false values - can they be removed?

JF: we can rely on computed role, inferred from the native semantics or from ARIA. Action and destination carries through the semantics. Do we want the semantics implied or inferred. We do need to be aware of conflict.

<Zakim> jf, you wanted to note that the values of True and False we reference are never used...

Matthew_Atkinson: my reason for suggesting combining the attributes is that I can't see a way of processing them separately;

Matthew_Atkinson: to justify separate attributes we need a use case that has a button that would legitimately be marked as a destination. If not, then it doesn't seem to make sense to have separate attributes.

<Matthew_Atkinson> Demo page: http://matatk.agrip.org.uk/personalization-semantics-explorations/demo.html

LisaSeemanKest: believe that attributes are processed differently; sometimes meaning changes, when devs. trying this some of the names were not clear enough. For example, sometimes people associated help with the region where help was provided. If it is a destination we may need to modify the name to help-page to be specific.

LisaSeemanKest: if purpose of the field is name, you might want "your name" as an automatic tooltip. But don't always want that tooltip

<jf> tabs do not open new pages

LisaSeemanKest: Tabs are an example - a tab to me is like a button but it also kind of opens a new page so the tab role could have a destination or it could have an action.

LisaSeemanKest; interaction model is a tab but context could be different

Matthew_Atkinson: if someone incorrectly puts help in a landmark region how do we indicate it is wrong? I have an example of how to handle incorrect attributes on my example page. I added a question page when something is asking for input, if it is on static text it uses an i (info) icon

Matthew_Atkinson: curious what do we do when it is implemented wrong? how to resolve? If author says its a button and it has a destination, how do we resolve? This is why implementing it more simply is a benefit. I can't see a button with a destination every being valid.

Janina: do we want hand coding vs applying principles. Don't want to see us overwriting on a case by case basis.

janina: when someone puts the incorrect marking on a landmark - this is a situation where these need to be cleaned up. I would prefer these be easier to implement on a large number of pages and frameworks and eliminate the necessity of "clean up"

JF: I like the simplicity of Matthew's proposal but am worried to see these attributes applied incorrectly, like help on a landmark

JF: expect people to make mistakes in the begining since this is new; believe we will also need an authoring patterns doc to address Lisa's concerns for author error

<Zakim> jf, you wanted to discuss author errors - we're too early to say whether or not it's really confusing

<jf> +1 linters (w3c validator) could catch errors with explicit attributes

LisaSeemanKest: it is easy to make an authoring tool that can pick up the mismatch errors (destination on a button); If people use the wrong attribute it is up the AT how they handle the errors; probably depends upon the target audience.

<jf> i.e. <a href="..." action="..."> == non-valid code

LisaSeemanKest: many of our target audience have trouble figuring things out, better to not add something to a page rather than add the wrong thing

LisaSeemanKest: we know AI isn't reliable enough, that is why we need to rely on assistive techs to understand their audience

<jf> it's a region, not a user-action

becky: what is the issue with putting help on a region?

<Matthew_Atkinson> +1 to Becky's point that the tooltip provided for a region would be made different to the tooltip for a button.

LisaSeemanKest: it makes it much harder to the AT to implement; If there is no role how does a UA handle this

<Zakim> jf, you wanted to note that these 3 attributes are all associated to user-actions

JF: concern that we are mixing contexts - the 3 attributes we are discussing now are interactive/actionable. We want to constrain to user actions or it will be too confusing

<LisaSeemanKest> 1- keep as is, 2 towards merging

<LisaSeemanKest> 0 no prefrence

2

<janina> 2

<LisaSeemanKest> 1 but cn live with iether

<Matthew_Atkinson> 2

<jf> vote: 1 (keep as is) with a note to dicsus conflict resolution

<Lionel_Wolberger> 0

<sharon> 0

<Roy> 0

<Matthew_Atkinson> *2 (but exploring merging it; I think we need to examine the use cases more)

janina: we don't want to make a decision in haste and regret it later.

LisaSeemanKest: we have already taken 2 years to investigate and made a decision

JF: Matthew asks 3 of our attributes feel very similar and can we merge them. think we need to focus on the outcomes. Matthew's examples actually used two icons to implement the full context - are we putting more effort to understand. Would prefer to put the burden on the author

<jf> @matthew - our goal is to encode semantics, not outcomes

Matthew_Atkinson: would appreciate feedback on the issues that are found on the examples I made. I don't see a difference on how the 3 attributes would be processed differently by an AT

<Zakim> Matthew_Atkinson, you wanted to clarify the question and focus on use the cases/examples

Matthew_Atkinson: regarding two icons that is to support the auto complete attribute can take two values; That is a separate issue we need to address.

<jf> +1 to addressing conflict resolution

<jf> I propose we defer to native semantics for all of our attributes

Matthew_Atkinson: believe we need to discuss conflict resolution no matter what direction we take with respect to single or multiple attributes

LisaSeemanKest: believe adoption of this spec will be small projects that have a big impact on a particular user group. Suspect these will be plugins to existing authoring tools rather than a completely new authoring tool.

<jf> Governmental pages too Lisa!

<jf> (e.g. US Department of Veterans Affairs)

LisaSeemanKest: We don't want to just cater to large organizations, if we do we will have missed the specialized audiences ( assistive living, rehab, etc.)

LisaSeemanKest: worried that the perfect is our enemy; we have been going around and around on this and if we keep doing that we will never finish. Appreciate Matthew

LisaSeemanKest: doing deep exploration and finding and raising issues;

<jf> easily 3 years

LisaSeemanKest: we discussed this for many years and made a decision, we can't keep reopening the decision.

<jf> +1 to Lisa's general point - are we really gaining a lot by revisiting this?

sharon: We have an open issues on action vs destination since 2018 and have never closed

janina: we need to be aware that the computed role will likely come up with spec gets reviewed by "higher up". We need to have our reasons well spelled out and understood

janina: education is also an early adopter; there is opportunity to become part of WCAG 3;

<Zakim> janina, you wanted to suggest this could be in wcag3

Lionel_Wolberger: both Matthew and I are recently recruited; I understand that there is fatigue but don't want to disregard Matthew's developer based comments

LisaSeemanKest: Please look over Matthew's comments and examples.

<LisaSeemanKest> https://github.com/w3c/personalization-semantics/wiki/Comparison-of-ways-to-use-vocabulary-in-content