See also: IRC log
<trackbot> Date: 28 January 2015
<janina> agenda: this
<scribe> scribenick: fesch
<Gottfried> Gottfried: JTC1 SC35 meeting last week in Copenhagen. Standards from the former JTC1 Special Working Group on Accessibility (SWG-A) are being transferred to SC35 now, including the user needs summary and standards inventory. Also, SC35 has requested to have WAI-ARIA standardized as an additional part of 13066.
last week minutes should show Gottfired sent regrets
RESOLUTION: publish minutes with above edit
reviewing actions
action 1557
action-1557
<trackbot> action-1557 -- Joanmarie Diggs to Investigate creating a w3c bugzilla bot -- due 2015-02-25 -- OPEN
<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1557
action-1558
<trackbot> action-1558 -- Fred Esch to Review compositing and blending level 1 http://www.w3.org/tr/2015/cr-compositing-1-20150113/ -- due 2015-01-28 -- OPEN
<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1558
<joanie> scribe: joanie
<fesch> lol
<fesch> scribe: fesch
make agenda item to clear up actions
<JF> suggest adding ACTION ITEMS cleanout to next week's agenda
<JF> www.w3.org/community/groups/#exploresemdata
color blind community formed in November....
color blind community has Michael C and Doug S.
action-1558
<trackbot> action-1558 -- Fred Esch to Review compositing and blending level 1 http://www.w3.org/tr/2015/cr-compositing-1-20150113/ -- due 2015-01-28 -- OPEN
<trackbot> https://www.w3.org/WAI/PF/Group/track/actions/1558
<joanie> scribe: joanie
FE: There are related notes in
the action.
... CSS allows colors with alpha values.
... If you have an alpha value of less than 1, the rendered
contrast is further reduced.
... But in the calculation, it doesn't use alpha.
... So if the colors only differ by alpha, the rendered colors
will be different, but the contrast calculated by the WAI
contrast formula will remain the same.
... This impacts how you test with color contrast.
JN: But you test the rendered contrast.
<JF> http://leaverou.github.io/contrast-ratio/
JF: The above tool allows for
contrast ratio testing.
... That tool reports back a range.
... An option is to grab an image and analyse that.
... But you have multiple issues: foreground/background,
opacity layers in the middle to impact contrast.
... So now there's three colors, not two, that you have to take
into account.
JN: What I do is remove the text
from the rendered display, take a screenshot of the background,
and use the most commonly occurring one.
... So it is possible to create tools to test for these
issues.
JF: Whether or not the W3C wants to look at creating such a tool, or wait for a third party, is a question.
JN: I thought the tool you were talking about takes a range. That, to me, is not all that useful.
JF: Yes, it's a useful hinting tool; not for compliance testing.
JS: I'm hearing one
recommendation from Fred's action, namely asking if the W3C
should host a color-contrast testing tool.
... I've not heard any suggested edits to the spec yet.
... I'm on advisory committees where people have done studies
where these sorts of issues have come up.
... Do we want to ask for the ability to kill transparency
entirely?
... Similarly, do we want to make it possible for users to
disable gradiants?
FE: I do suggest we change the
spec.
... So people cannot just say, "I did the check and calculated
it."
... Also, compliance tools may use the spec.
... Floating dialogs that pop up and are semi-opaque, are
another problem.
... We're not going to tell CSS they cannot have transparent
colors.
<Zakim> janina, you wanted to ask about gradients and about disallowing transparency/gradients
<Zakim> JF, you wanted to say that suggesting "Killing alpha-transparency and gradients is a fools errand"
JF: Venturing down the path is a losing battle.
JS: I was proposing it would be users who could disable it.
JF: (Stands by his opinion)
CS: To Fred, are you talking about the 4.5 contrast ratio?
FE: Yes, and the 7.5 one.
CS: What's wrong with it?
FE: It does not use an alpha value.
CS: This equation is really popular, so I think we need to update it to consider alpha.
JS: As well as multiple levels, based on what John is saying.
JN: I don't think it's worth it. We're not going to be able to cover all possible scenarios.
<fesch> the formula does not use alpha values, it needs to
JN: You should not just use the colors stated in the background.
JS: I'm not talking about author
requirements, or banning gradiants.
... They're cool. There are better ways to use them. There's
the possibility of testing tools added to W3C.
... I'm just asking if it's reasonable or possible for enabling
the *user* to disable the gradiants and transparency.
CS: In response to James'
comment, the formula is what gets used.
... Text may be skimmed around. So I don't think notes are
going to solve this problem.
... Also, I think some of the gradiants and things go away in
Windows's high contrast so it's technically feasible.
JF: I tend to agree with
Cynthia.
... There's already algorithms for alpha blending.
... So the only real change would be to add the background and
middleground and do the blending to arrive at the final
color.
... Right now we need a tool to accomplish what we're doing
visually.
JN: If you could do a formula,
I'd be for it.
... You don't know what's behind it. The background could be an
image.
... The only thing you can rely upon is what's rendered.
CS: Good point.
... The formula was based on research Greg V directed. Perhaps
we could check with him to see who worked on that.
... If so, maybe we can convince him to do an update.
JS: I will follow up on that.
JN: The formula might be in the normative part of WCAG.
JS: We'll try to avoid changing anything normative.
CS: I think you're right, James. I think it normative.
FE: The formula is very important. If nothing else, for automated testing tools, at least to flag possible issues.
<JF> ACTION: Janina to contact Greb V re: research on color contrast calculations due Feb. 11 [recorded in http://www.w3.org/2015/01/28-pf-minutes.html#action01]
<trackbot> Created ACTION-1571 - Contact greb v re: research on color contrast calculations due feb. 11 [on Janina Sajka - due 2015-02-04].
FE: So we really need to make sure the formula addresses this.
JS: I think we all agree manual testing is no longer good enough.
<JF> ACTION-1571 due 2015-02-11
<trackbot> Set ACTION-1571 Contact greb v re: research on color contrast calculations due feb. 11 due date to 2015-02-11.
JS: We know that background
images matter. If the browsers can provide a means for us to
take that into account it would be useful.
... We seem to agree that the W3C should consider hosting a
testing tool for this.
CS: I'd would caution on the W3C creating a tester. There are already plenty of checkers out there.
<JF> ACTION: Cynthia to look at the possibility of using Browser code for alpha transparency checking due 2014-02-11 [recorded in http://www.w3.org/2015/01/28-pf-minutes.html#action02]
<trackbot> Created ACTION-1572 - Look at the possibility of using browser code for alpha transparency checking due 2014-02-11 [on Cynthia Shelly - due 2015-02-04].
JS: I'm still not clear if we need to say anything to CSS.
FE: I don't think there's
anything for us to say to CSS.
... I think we need to be sure our standards keep up with
reality.
JF: This really should include
the WCAG technique folks.
... I don't think we have techniques for this yet.
... I'm not sure if Andrew, Josh, and others are looking at
this.
JN: We're not actively working on
this at the moment. But yes we should raise it.
... A more pressing issue is how do you determine what the
background color is when it varies?
(Group discusses "bad pixels")
JN: I have an algorithm, but I don't know if it's sufficient.
CS: Perhaps we can talk to Greg about this.
FE: CSS has formulas for what
comes out (as RGBA value).
... So you might be able to calculate the contrast from
that.
JS: Maybe the spec should include language about insufficient contrast.
CS: That sounds like something we could add to the formula.
FE: The compositing math is there.
JN: If you have two or more flat colors, you can do it from the algorithm, that works. But that's not what happens in the real world.
<JF> +1 to James' point
CS: I think there's two pieces: Determining how many "bad pixels" is too many; transparency of two solid colors, combined together.
JS: Status of this spec?
FE: I thought this week it was
going into CR.
... This is reality. We cannot do anything with the CSS spec,
it's a matter of what we need to do.
... I don't think we want to recommend anything added to the
CSS spec.
CS: Should we recommend referencing WCAG?
JS: We need to have a follow-up
discussion on this: To what extent do we want to push
this?
... WCAG is presumed to apply to all specs.
... Let's touch on this topic next week.
... Thanks to Fred for the review!
JS: We're moving to publishing working drafts in February when Michael returns.
<fesch> rssagent, make loogs
<fesch> rssagent, make logs
<fesch> rssagent, draft minutes