W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 2017

Re: Color Test: A formal proposal

From: David MacDonald <david@can-adapt.com>
Date: Sat, 1 Apr 2017 19:07:26 -0400
Message-ID: <CAAdDpDbeyqCdThgwMvWBRBdb+ha_hZQ9ER3Z3yue_XWGvturaw@mail.gmail.com>
To: Gregg C Vanderheiden <greggvan@umd.edu>
Cc: Dick <wayneedick@gmail.com>, WCAG <w3c-wai-gl@w3.org>, public-low-vision-a11y-tf <public-low-vision-a11y-tf@w3.org>
This test is saying the dev only has to test one colour but is responsible
for all 256,000,000. The auditor can fail him on things he didn't test.

It can't work like that.

Then the developer could get sued or fired for not meeting the WCAG even
though they did everything they needed.

The way it should work is that the dev would to say in his statement of
conformance, the values tested... just like "we tested with JAWS 18, on WIN
and IE 11" and the auditor tests that. Now if the auditor decides to check
something else and say, "hey I noticed this didn't work" that is a best
practice statement. I do that all the time with devs

Having said that, if one colour can be overridden successfully many others
will...

On Apr 1, 2017 3:58 PM, "Gregg C Vanderheiden" <greggvan@umd.edu> wrote:

> Just FYI
>
> Technically — we don’t have any such things as “formal tests”  except for
> TECHNIQUES.
>
>
> This can’t be a formal test unless the SC says that you must do exactly
> this - or rather  the SC must say  “  Content passes the following test"
>
> you put it forward as an informal test — but the SC is the only criterion
> for passing the SC.  (that is what its name means—  success criterion.
>
> The WG COULD propose the test as a ‘sufficient’ test of the SC.   That is
> — if you pass, you pass.
> But you cannot say that if you fail you fail unless the SC says this
> specifically.
>
> Gregg
>
> Gregg C Vanderheiden
> greggvan@umd.edu
>
>
>
> > On Apr 1, 2017, at 2:49 PM, Wayne Dick <wayneedick@gmail.com> wrote:
> >
> > I proposed this color test.
> > It should work.
> > The colors are selected randomly so that they support a 4.5:1 ratio.
> > This test should be sufficient.
> > It tests two random color choices (one dark one light).
> > The combination is most likely a mud color (light or dark).
> > The test looks at dark on light and light on dark.
> > It is significant that !important is left off the first test.
> > It should be run  twice, without and with !important.
> > The non-important will flush out element level style.
> > The important will flush real erroneous cases.
> >
> > Look for colors that do not change.
> > Loss of functionality, images disappear, icons dispensary
> >
> > If colors do not change add-in background-image: none.
> >
> > Pay attention to borders and padding. These may also need to be
> specified.
> >
> > I would like to put this forward as a formal test.
> >
> > Wayne
> >
> >
>
>
>
Received on Saturday, 1 April 2017 23:07:59 UTC

This archive was generated by hypermail 2.3.1 : Saturday, 1 April 2017 23:08:00 UTC