Issue 433 revise or drop F87: Failure of Success Criterion 1.3.1

<shawn> scribe: shawn

<laura> Issue 433 revise or drop F87: Failure of Success Criterion 1.3.1

<laura> https://github.com/w3c/wcag/issues/433

<scribe> scribe: Jim

<laura> F87: https://www.w3.org/TR/WCAG20-TECHS/F87.html

<laura> AGWG Minutes: https://www.w3.org/2018/07/31-ag-minutes.html#item09

<AllanJ> jim: assume there is a way with css to change fonts everywhere except ::before and ::after

<AllanJ> jim: basic issue is css :content can be used in the accessible name calculation. does :content have an impact on LV folks

<AllanJ> laura: technique icon font with role=image

<laura> https://www.w3.org/WAI/GL/wiki/Providing_a_Semantically_Identified_Icon_Font_with_role%3Dimg

<AllanJ> wayne: created css to change fonts and ignore role=image

<AllanJ> ... there are data-* attribute that we could hook into. may need an aria-psuedo element.

<AllanJ> jon: if role=image, it must be applied to whole element. we keep the font icon but can't change the font on the parent element

<AllanJ> ... perhaps some :not css

<AllanJ> ... if you use wayne anchor css befor/after []

<AllanJ> jim: is there a test page to jump off from

<AllanJ> https://codepen.io/Wildebrew/pen/yqNWXG

<AllanJ> jim: if we use example and apply wayne's font css what happens

<AllanJ> wayne: can use the code in technique to demonstrate issues.

<AllanJ> wayne: user stylesheets are critical. before/after break too many things. have eliminated them.

<AllanJ> jon: updated issue 433, when pseudo elements used for content, browser search does not find the content.

<AllanJ> jim: any bugs filed on browsers for 'find' not finding css-content

<AllanJ> jon: would like css to be separate from content, but they are totally entwined.

<AllanJ> ... need better browser support for interacting with pseudo elements

<AllanJ> wayne: Find is a critical issue.

<AllanJ> jon: people use pseudo elements for many things. some cases are not an issues, others are.

<AllanJ> ... need to scope how it should be used. needs to be qualified

<AllanJ> wayne: feels authors will ignore

<AllanJ> wayne: need mechanisms to process this information. AT cannot get at the information

<AllanJ> laura: need to add info on User Agent support section.

<laura> User Agent Support Notes for F87: https://www.w3.org/TR/WCAG20-TECHS/F87.html

<AllanJ> wayne: add browser "FIND" not supported for finding pseudo content

<AllanJ> jon: applied css with * to change pseudo elements - it works.

<AllanJ> ... we need to document the vagries

<AllanJ> jim: how do we want to handle this - list, wiki, issue

<AllanJ> wayne: add [problem] in the issue message

<AllanJ> wayne: will share css

<AllanJ> jon: collision between user styles pseudo and author pseudo

<AllanJ> wayne: pseudo element have no structure, they are just flat

<AllanJ> jon: css doesn't have syntax to say replace fonts that are not pseudo elements.

<AllanJ> Jim: I can write css-tricks to see if they can find a way to make it work.

<AllanJ> wayne: need to be deterministic. always be able to find it.

<AllanJ> ... with frameworks, etc. the ids are always changing, hard to keep a consistent user style

<AllanJ> jon: can't put a pseudo inside a not.

<AllanJ> jon: if you change all the fonts, no way to find the original font the author.

<AllanJ> wayne: css does not have a "don't inherit" for anyelement.

<AllanJ> ... change the parent, then subelements also change

<AllanJ> jim: do pseudo elements cascade with user and author. can both appear at the same time.

<AllanJ> wayne: pseudo elements do not have enough structure (data) within them to support 1.3.1

<AllanJ> wayne: thinks we need to keep F87, will create an example.

<AllanJ> jon: used STYLUS and !important to overwrite some pseudo elements, it failed

<AllanJ> ... stylus would not obverride the speudo or concatenate.

<AllanJ> jim: F87, ACname calc - may change because of custom stylesheet.

<AllanJ> aria-group controls AccNameCalc.

<AllanJ> jon: icon font - can't put aria-label on icon font, must put aria on the parent element

<laura> A low vision user may not be using a screen reader.

<AllanJ> jon: fed gov still using IE. voiceover on IOS separates all elements, so it stops at each element

<AllanJ> first element, spans, etc. - more of an AT non-parsing issue

<AllanJ> jason: will test in Edge.

<AllanJ> wayne: Edge, the only browser that reads epub automatically. Nice.

<AllanJ> jason: try to stay very compliant with WCAG, ARIA, screen readers, narrator

<laura> andrew’s example: http://awkawk.github.io/f87.html

<AllanJ> process survey for wcag2.1

<laura> Process feed back suvey: https://www.w3.org/2002/09/wbs/35422/processfeedback/

<AllanJ> Laura: do we drop F87

<AllanJ> wayne: no, Find is a huge issue. May need to scope.

<AllanJ> jon: use find for keyboard navigation. use find, jump to part of page, then tab to move to item

<AllanJ> jon: pseudo elements are not removed in high contrast mode in IE/Edge - a reason to use them

<AllanJ> laura: need to test first to find conflicts.

<AllanJ> jon: failing a failure technique, just have to pass a different way.

<AllanJ> jon: have F3 (background images), for decorative images. Low vision folks need a way to remove decorative backgound images. F3 and F87 are related.

<AllanJ> wayne: +1, perhaps create new techniques for inserting or generating content in a LV friendly way.

<AllanJ> wayne: any objections to list counters. can't apply css to them

<AllanJ> jon: some WAI document used pseudo elements to indicate heading numbers, but could not search for them with 'find'.

This is scribe.perl Revision: 1.152  of Date: 2017/02/06 11:04:15  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Present: jon_avila Laura Jim Jason Shawn Wayne
Found Scribe: shawn
Inferring ScribeNick: shawn
Found Scribe: Jim

WARNING: 1 scribe lines found (out of 95 total lines.)
Are you sure you specified a correct ScribeNick?

Scribes: shawn, Jim
Found Date: 02 Aug 2018
People with action items: 

WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)