W3C

– DRAFT –
Personalization Task Force

29 March 2021

Attendees

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

Meeting minutes

agenda, take up item 1

PR related to Action 79 and the [email sent to list 22 Mar 2021] (John) https://lists.w3.org/Archives/Public/public-personalization-tf/2021Mar/0025.html

<JF> https://github.com/w3c/personalization-semantics/pull/176

JF: Done and in Git.
… Recap: term+term with no spaces.

becky: merge done. waiting for the build now.

becky: let's recap the issue.

JF: Browsers store many tokens (name, address, etc) and they are generally free-form.
… But i18n seems to think country and language will be encoded as per ISO standards.
… yet they acknowledge that country could be a text string, and not a country code

janina: This is not the group for this discussion.

<JF> https://github.com/whatwg/html/issues/6519#issuecomment-806578195

sharon: There are three i18n issues, which have been lingering.

janina: Happy to close this one for now.

sharon: let's return to this later

investigate Research @rel versus @action and @destination and report back to tf (Action 78 - John)

i18n issues 144, 145, 175 - https://github.com/w3c/personalization-semantics/issues/

<becky> https://github.com/w3c/personalization-semantics/issues/144

JF: i18n does not understand what we are trying to do here.
… we are creating a machine readable taxonomy
… stored values will be echoed back to users, and the machine is agnostic to the content of the tokens
… I could put my language as "Pig Latin" in the browser, even though this is not a real language

janina: Taking this on here, in our spec, is 'above our pay grade'
… thanks JF for filing the issue in their group
… I ask the group not to deep dive into it.

becky: This is a little bit different. We rely on the host language, e.g. HTML, and the assistive technology will adhere to that language in its representation.
… that may be what hthey are asking for.

JF: We are providing the metadata

becky: we do not provide the metadata, it's elsewhere in the stack. We rely on the hosting document to specify the language and the assistive technology implementing our specification will adhere to that

JF: We are not creating content here, yet i18n thinks we are

becky: refining the above
… we do not provide the language and direction metadata. We rely on the hosting document to specify the language and metadata.
… the -personalization specification is expected to adhere to that information.

JF: Proposing: the output is from the user's submitted content.

janina: Developers will run checks on it, if this content is useful (e.g., will be used by the browser)
… meaningless stuff sont be used

JF: Disagree

JF: The format of the output can be meaningless because the machine is GIGO

janina: Why do we need to state it

JF: Response values can all be meaningless. Pig latin. Klingon.

janina: I maintain, we do not need to cite this in the response.

janina: ... you brought it up to HTML

<Zakim> JF, you wanted to note we ARE inventing something new

JF: Token values for autocomplete are being repurposing in our specification. We are inventing atsymbol, which takes a list of token values.
… we take the tokens from autocomplete, and things that autocomplete cannot be applied to
… autocomplete cannot be applied to a fieldset
… or a div
… so we are inventing something new.
… then we should explicitly state that this new invention is not constrained by the constraints that autocomplete is experiencing in the HTML specification

becky: i18n was saying, you have these 'symbol things', but you didn't say language and direction
… this one can be seperately set, and is not tied into 'purpose'

sharon: aphillips response is ambiguous. Can we close this or not?

JF: I read, "Thanks for the update." as this can be closed.

JF: replace 'user agent' for 'assistive technology' in the above

janina: (AT, browser extension, etc.)

<JF> 4 days ago: https://github.com/whatwg/html/issues/6519#issuecomment-806578195

becky: the user agent seems resposnible for this language field. e.g., chrome, firefox

JF: The browser is supposed to echo back whatever value the user has typed. i18n seems to expect that the browser will restrict itself to valid BCP47 values.
… i18n does not specify who is responsible for translating it from free text to bcp47
… my research into success criteria 1.3.5 revealed that no user agent is supporting 'language'
… I contend that this is due to the onerous requirement that user agents restrict input to bcp47
… there are over 300 options, so a drop down is out of the question

janina: propose, we publish with a note regarding bcp47 guidance
… after waiting one more week

PROPOSAL: Personalization notes that there is ambiguity regarding the formatting of language: i18n requested adherence to BCP47 but there is a strong opinion that such adherence would restrict user activity unnecessarily.

becky: PROPOSE: The user doesn't know BCP47

CharlesL: could link to the issue, in this note

becky: there is potential ambiguity in how this is specified, and currently there is disparate implementations in user-agents

Autocomplete does depend on user-agents for the user-supplied values

CharlesL: might add the link to the WHAT-WG, https://github.com/whatwg/html/issues/6519

+1

becky: validation is limited, and not necessarily assigned to the user-agent

<JF> Our spec should not be responsible for form validation

+1 to that

<JF> +1 to Charles

CharlesL: Users want to see human readable terms for language. Who wants to see BCP47 values in their form fields?

<janina> https://www.evertype.com/standards/iso639/sgn.html

JF: Current practice shows, validation is not done

becky: i18n says we should say our spec should call this out; we can say, HTML called it out. We can simply say purpose is language.

<JF> So we DON'T agree with i18n

becky: The word 'language' here is not a BCP47 reference, it is a request for a purpose value

JF: The spec says nothing about purpose

janina: That's not good logic, since we are discussing language in general
… our interest is about more than 'purpose', since you must then ask, 'purpose of what'

<JF> the3 spec says NOTHING about @purpose, it speaks to @autocomplete - and there is a bug filed there against the language value

sharon: Decided, we will wait another week and see what / if i18n responds

<JF> +1 to waiting another week

janina: Scheduling: skip next week, Easter Monday>

Resolution: We taking Easter Monday off.

zakim: end meeting

Summary of resolutions

  1. We taking Easter Monday off.
Minutes manually created (not a transcript), formatted by scribe.perl version 127 (Wed Dec 30 17:39:58 2020 UTC).

Diagnostics

Succeeded: s/+1 to Chalres/+1 to Charles

Maybe present: zakim