W3C

– DRAFT –
Personalization Task Force Teleconference

10 May 2021

Attendees

Present
CharlesL, janina, JF, Lionel_Wolberger, LisaSeemanKest, Roy, sharon
Regrets
becky
Chair
-
Scribe
charlesl

Meeting minutes

<LisaSeemanKest> agenda+ True/False values (John) - https://lists.w3.org/Archives/Public/public-personalization

<LisaSeemanKest> Conflict resolution (John) -

<LisaSeemanKest> https://lists.w3.org/Archives/Public/public-personalization

<LisaSeemanKest> -tf/2021Apr/0032.html

<LisaSeemanKest> present

<LisaSeemanKest> next item

<JF> https://www.w3.org/TR/personalization-semantics-1.0/

JF: true/false was legacy, in the explainer which is not normative. We are not using true/false so we can just remove it from the explainer.

<JF> https://www.w3.org/TR/personalization-semantics-1.0/#propcharacteristic_value

JF: , not used in any of the modules, so we can just take them out.
… , we either defined either a string or a specific set of token values, but nothing used true/false. In this section in the explainer we can take out true/false and true/false undefined and maybe URI as well.

<LisaSeemanKest> +1

+1

<Lionel_Wolberger> +1

<Matthew_Atkinson> +1

<janina> +1

<sharon> +1

Action: JF to clean up Explainer to remove values (T/F, etc.)

<trackbot> Created ACTION-82 - Clean up explainer to remove values (t/f, etc.) [on John Foliot - due 2021-05-17].

<LisaSeemanKest> next item

Resolution: remove true/false from the explainer

<LisaSeemanKest> next item

JF: we may have already answered this over the past 2 weeks. For the first 3 attributes (action, destination, purpose) there are times when those 3 were to be paired with native role semantics. What happens when action is not on a button, but links should have destination attributes.

<LisaSeemanKest> https://lists.w3.org/Archives/Public/public-personalization-tf/2021May/0007.html

<JF> <a href="" role="button" action="opens-in-page-dialog" destination="help">

JF: , role of button then takes on the action. we only augment the semantics. we need to tighten that up.
… , we don't change the semantics of the role of the element.

<Matthew_Atkinson> +1 to us augmenting (not overriding) semantics

Lisa: I suspect it should not change it, because it could break thinks. should say algorithm checking should issue a warning, not an error. for validators.

JF: I agree, Mike Smith is the maintainer of the validators right now. he could adopt the rules as required. we also at one time thought our would be more aligned with ARIA which does change the semantics. However I don't think this is the case for our spec.

True/False values (John) - https://lists.w3.org/Archives/Public/public-personalization

<Roy> JF: true/false was legacy, in the explainer which is not normative. We are not using true/false so we can just remove it from the explainer.

<Roy> [2021-05-10 22:04] <JF> https://www.w3.org/TR/personalization-semantics-1.0/#propcharacteristic_value

<Roy> [2021-05-10 22:04] <CharlesL> …, not used in any of the modules, so we can just take them out.

<Roy> [2021-05-10 22:06] <CharlesL> …, we either defined either a string or a specific set of token values, but nothing used true/false. In this section in the explainer we can take out true/false and true/false undefined and maybe URI as well.

<Roy> [2021-05-10 22:06] <LisaSeemanKest> +1

<Roy> [2021-05-10 22:06] <CharlesL> +1

<Roy> [2021-05-10 22:06] <Lionel_Wolberger> +1

<Roy> [2021-05-10 22:06] <Matthew_Atkinson> +1

<Roy> [2021-05-10 22:06] <janina> +1

<Roy> [2021-05-10 22:06] <sharon> +1

<Roy> [2021-05-10 22:06] <JF> ACTION: JF to clean up Explainer to remove values (T/F, etc.)

<Roy> [2021-05-10 22:06] * trackbot is creating a new ACTION.

<Roy> [2021-05-10 22:06] * RRSAgent records action 1

<Roy> [2021-05-10 22:06] <trackbot> Created ACTION-82 - Clean up explainer to remove values (t/f, etc.) [on John Foliot - due 2021-05-17].

<Roy> [2021-05-10 22:06] <LisaSeemanKest> next item

<Roy> [2021-05-10 22:06] * Zakim sees a speaker queue remaining and respectfully declines to close this agendum

<Roy> [2021-05-10 22:06] <LisaSeemanKest> q?

<Roy> [2021-05-10 22:06] * Zakim sees Matthew_Atkinson on the speaker queue

<Roy> [2021-05-10 22:06] <CharlesL> Resolved: remove true/false from the explainer

<Roy> [2021-05-10 22:06] <LisaSeemanKest> ack m

<Roy> [2021-05-10 22:06] * Zakim sees no one on the speaker queue

<Roy> [2021-05-10 22:06] <LisaSeemanKest> next item

<Roy> [2021-05-10 22:06] * Zakim thinks agendum 2 -- Conflict resolution (John) - https://lists.w3.org/Archives/Public/public-personalization-tf/2021Apr/0032.html -- taken up [from LisaSeemanKest]

<Roy> [2021-05-10 22:08] <LisaSeemanKest> agenda+ matts othe emails

<Roy> [2021-05-10 22:08] * Zakim notes agendum 5 added

<Roy> [2021-05-10 22:09] <CharlesL> JF: we may have already answered this over the past 2 weeks. For the first 3 attributes (action, destination, purpose) there are times when those 3 were to be paired with native role semantics. What happens when action is not on a button, but links should have destination attributes.

<Roy> [2021-05-10 22:09] <LisaSeemanKest> https://lists.w3.org/Archives/Public/public-personalization-tf/2021May/0007.html

<Roy> [2021-05-10 22:09] <JF> <a href="" role="button" action="opens-in-page-dialog" destination="help">

<Roy> [2021-05-10 22:10] <CharlesL> …, role of button then takes on the action. we only augment the semantics. we need to tighten that up.

<Roy> [2021-05-10 22:10] <CharlesL> …, we don't change the semantics of the role of the element.

<Roy> [2021-05-10 22:10] <Matthew_Atkinson> +1 to us augmenting (not overriding) semantics

<Roy> [2021-05-10 22:11] <CharlesL> Lisa: I suspect it should not change it, because it could break thinks. should say algorithm checking should issue a warning, not an error. for validators.

<Roy> [2021-05-10 22:13] <CharlesL> JF: I agree, Mike Smith is the maintainer of the validators right now. he could adopt the rules as required. we also at one time thought our would be more aligned with ARIA which does change the semantics. However I don't think this is the case for our spec.

<LisaSeemanKest> note the server went down so we shifted to zoom chat. charles is coping it back

This specification, suggests but does not overide native semantics. In the case of conflict between implied sematinics validation algorims should issue a warning, not an error.

Me to Everyone (7:20 AM)

Matthew: semantics from the a11y tree. Implied semantics could be confused with implicit roles

Lisa Seeman to Everyone (7:20 AM)

This specification suggests, but does not overide, native semantics and does map to accessibilty tree. In the case of conflict between implied sematinics and native roles, validation algorims should issue a warning, not an error.

JF to Everyone (7:20 AM)

The attributes and values in this specification do not overide the semantics found int he Accessibility Tree, but rather augment them. In the case of conflict betbween implied sematinics validation algorims should issue a warning, not an error.

The attributes and values in this specification do not overide the semantics exposed by the Accessibility Tree, but rather augment them. In the case of conflict between implied sematinics, validation algorithms should issue a warning not an error.

Me to Everyone (7:23 AM)

Matthew: action or destination that author got it wrong, but some UA might just do it anyways. If we made it hard and valid but could see it.

Me to Everyone (7:23 AM)

Lisa: they don’t map to the a11y tree / change the a11y tree, and when it conflicts should be an error.

Matthew: maybe leave out the conflict resolution.

JF to Everyone (7:24 AM)

The attributes and values in this specification do not over-ride the semantics exposed by the Accessibility Tree, but rather augment them.

Lisa Seeman to Everyone (7:25 AM)

who likes that wording?

+1

Me to Everyone (7:25 AM)

+1

JF: the broadest meaning of semantics applied at the element level.

JF: making it machine readable.

Me to Everyone (7:26 AM)

Matthew: should it be exposed in the a11y Tree or by?

JF: I would say it should be injected into the a11y tree. What ever the agues are should be nodes in the a11y only informational.

Matthew: agreed.

Roy Ran to Everyone (7:28 AM)

We could back IRC now

Me to Everyone (7:28 AM)

JF: Steve Faulkner might be able to help us with the wording.

Lisa Seeman to Everyone (7:31 AM)

matt will ask steve. but ediroal

wording subject to edorial change +1

Matthew Atkinson to Everyone (7:31 AM)

+1

Sharon Snider to Everyone (7:31 AM)

+1

Lisa Seeman to Everyone (7:31 AM)

anyone object

Lionel Wolberger to Everyone (7:32 AM)

+1

Matthew Atkinson to Everyone (7:33 AM)

I asked Steve if an element’s semantics are exposed in, or by, the accessibility tree, and he said that they are "exposed in the accessibility tree" :-).

Lisa Seeman to Everyone (7:33 AM)

thanks

Me to Everyone (7:33 AM)

Lisa: should validators throw a warning if the the semantics does not match the role?

Me to Everyone (7:34 AM)

Matthew: I am not sure if I object. Thinking fwd. when an author gets it wrong what the UA/Extension does, wondering how that affects the spec.

Lisa: validation algorithms that check code

JF: validators are good for validating the code by the author.

Me to Everyone (7:36 AM)

Lisa: lets decide on the validator should posts warnings, not an error, but that leads to the next discussion what happens if it still is wrong.

JF: but we then also need what UA should do.

Me to Everyone (7:37 AM)

Lisa: I disagree that we should tell the UA what to do if there is an error.

JF to Everyone (7:37 AM)

<input type="text" action="help">

Conflict resolution (John) - https://lists.w3.org/Archives/Public/public-personalization-tf/2021Apr/0032.html

<JF> The attributes and values in this specification do not overide the semantics exposed by the Accessibility Tree, but rather augment them. In the case of conflict between an element's semantics and the attribute values, validation algorithms should issue a warning but not an error.

<LisaSeemanKest> In the case of conflict between implied sematinics validation alg\orims should issue a warning, not an error.

Janina: so the value of issuing the warning or error, whoever is authoring the content they are potentially doing something wrong.

Lisa: they gotten something wrong that we are not requiring this, which is why this is a warning.
… , will help people use our specification, this will be a support to ARIA as well.
… , anyone object to this?

<JF> no objection so far

<LisaSeemanKest> put it in?

<LisaSeemanKest> +1

<janina> +1

<JF> +1

<sharon> +1

<Matthew_Atkinson> +1 (but Steve said it's "exposed in the accessibility tree" re the editorial question)

+1

Action: JF to add The attributes and values in this specification do not overide the semantics exposed by the Accessibility Tree, but rather augment them. In the case of conflict between an element's semantics and the attribute values, validation algorithms should issue a warning but not an error.

<trackbot> Created ACTION-83 - Add the attributes and values in this specification do not overide the semantics exposed in the accessibility tree, but rather augment them. in the case of conflict between an element's semantics and the attribute values, validation algorithms should issue a warning but not an error. [on John Foliot - due 2021-05-17].

Lisa: do we want to suggest how UA's handle this.

<JF> <input type="text" action="help">

Lisa: , they may want to add help overlays, or change it to something more ambiguous.

JF: I disagree, then my browsers tries to put an action icon on a text field which will be confusing. I think if its wrong it *MUST* be ignored.

<Matthew_Atkinson> What about this example (less straightforward maybe?): <a href="/help.html" role="button" destination="help">Help</a>

Janina: I would rather let UA do what they want, I don't see any direct benefit to it and allows for healthy competition.

<JF> <a href="/help.html" role="button" purpose="help">Help</a>

Lisa: Depending on what the UA is doing, maybe just putting on a Help icon.

JF: that is broken,

Matthew: if we simplifier things, we may end up with fewer attributes if we allow the UA to do what they want.
… , yes thats invalid,

JF: what should a browser do with that clearly wrong code?

Matthew: thats a good question, we will see what happens how they use it, we could give them some hints. I don't know the answer to the question. we want to see what people do with it and limit what they should do with it.
… , we we say its an error then thats it but since its a warning we will see what the UA do with it.

Lisa: make a proposal.

<LisaSeemanKest> may ignore

Lisa: the UA "MAY" ignore our semantics if its in conflict.

JF: for the validator is it a MAY/SHOULD or MUST?
… , issue a warning.

Lisa: I would go for a "SHOULD"

<LisaSeemanKest> +1 to should

<JF> 1) Validator SHOULD issue a Warning, User-Agent MAY ignore the code error

Janina: I agree for now *SHOULD* issue a warning, we are very young spec, we need to see creativity by the community being overly restrictive.

Lisa: I think we remove the UA only a SHOULD for Validators.

<LisaSeemanKest> any objetion to ignoring user agent recomendation

<JF> 0

<LisaSeemanKest> +1 to ignore

0

<sharon> 0

<LisaSeemanKest> janina stongly in agreements to ignoiring the user agent recomendation on conflict recomendation

<LisaSeemanKest> 4

<Matthew_Atkinson> 0 (maybe somewhere betwen 0.1 and 0.5, need to think, but doesn't change the result, and very interesting stuff! :-))

<LisaSeemanKest> siren

Summary of action items

  1. JF to clean up Explainer to remove values (T/F, etc.)
  2. JF to add The attributes and values in this specification do not overide the semantics exposed by the Accessibility Tree, but rather augment them. In the case of conflict between an element's semantics and the attribute values, validation algorithms should issue a warning but not an error.

Summary of resolutions

  1. remove true/false from the explainer
Minutes manually created (not a transcript), formatted by scribe.perl version 131 (Sat Apr 24 15:23:43 2021 UTC).

Diagnostics

Succeeded: s/Fawkner/Faulkner/

Succeeded: s/exposed by the/exposed in the

Maybe present: Lisa, Matthew