W3C home > Mailing lists > Public > public-css-testsuite@w3.org > December 2009

Re: Scrollbars, olive color, maximum viewport width

From: Gérard Talbot <css21testsuite@gtalbot.org>
Date: Mon, 21 Dec 2009 11:28:51 -0800
Message-ID: <891cd9d998a3041b71286f9629584758.squirrel@cp3.shieldhost.com>
To: public-css-testsuite@w3.org
Cc: "fantasai" <fantasai.lists@inkedblade.net>, "James Hopkins" <james@idreamincode.co.uk>

> Gérard Talbot wrote:
>>>> Ideally, the tests should not rely on the Uncommon Assumptions, and
>>>> if they
>>>> do, that should be indicated in the test.
>> Good enough then. I will add such PREREQUISITE in/for each and all
>> tests
>> requiring it.

I have now added the following PREREQUISITE text in 12 tests:

PREREQUISITE: User agent needs to support scrollbars as the scrolling
mechanism. If it does not, then this test case does not apply to this
user agent.

[Addendum: I now think that such prerequisite text in some of those
tests may not be required anyway ... even for UAs which do not create a
visible scrolling mechanism on block-level elements. UAs which do not
create a visible scrolling mechanims on block-level elements could still
pass or fail those max-* and min-* tests.]

> See also below.
>> Well then why not
>> <meta name="prerequisite" content="User agent needs to support
>> scrollbars as the scrolling mechanism. If it does not, then this test
>> case does not apply to this user agent.">
>> or
>> even add "scrollbars" as another possible token in the set of
>> "Requirement Flags"
>> http://wiki.csswg.org/test/css2.1/format#requirement-flags
>> <meta name="assumption" content="..."> is another possibility and is
>> ok
>> with me.
> The metadata in the test isn't visible to the tester, so it's more
> metadata isn't always a good solution. In most cases, a test for
> scrollbar behavior will pass on a UA that scrolls but has no
> scrollbars -- and ideally tests should be designed that way. In
> cases where the test will fail when scrollbars are not present,
> then the tester needs to be alerted so that he doesn't mark the
> test as failing when it simply doesn't apply.
> A number of the existing flags are there so that a tester can filter
> out tests that don't apply. There's some cognitive overhead in
> having a flag, so I'd prefer to avoid adding a flag and using the
> descriptive text instead if there's only one or two tests in the
> suite that might require it.

I agree with your analysis: such metadata would not be visible to
"mother/husband/roommate/brother/bus driver" testers and it may imply a
cognitive overhead.

>>>> That's probably a good idea. Do you have some recommendations?
>>> I second this. I would suggest standardizing the term, 'scrolling
>>> mechanism' as I believe it describes in a concise manner, the
>>> mechanism that is being described, and I also believe it's a term
>>> what
>>> would be clearly understood by reviewers. This term is also flexible
>>> enough to allow it to be extended to cover axis-specific scrolling,
>>> and active/inactive mechanisms; for example, "active scrolling
>>> mechanism along the x-axis", "inactive scrolling mechanism along the
>>> y-
>>> axis", etc.
>> What other scrolling mechanism should we assume (or expect) can be
>> tested by the CSS 2.1 test suite? And how can we test this? The CSS
>> 2.1
>> test suite mentions "panner" [1] as a possible scrolling mechanism.
> We don't have to test the scrolling mechanism itself, we just need to
> test that scrolling works and the rendering behaves as CSS2.1 requires.

Here's the real issue. In about 10 tests or so, the presence of
scrollbar(s) becomes part of the test itself as it can affect the
visible area of content width and/or content height of a block. The CSS
2.1 spec requirement goes like this:
Any space taken up by the scrollbars should be taken out of (subtracted
from the dimensions of) the containing block formed by the element with
the scrollbars.




(some Firefox 3.7 alpha build will render scrollbars for unknown

(I believe - but I could be wrong here - that *both* scrollbars presence
should be a decisive, determining pass/fail factor in such test).

> Most tests should be scrolling-mechanism-agnostic; so we need standard
> wording that is also scrolling-mechanism-agnostic.

I believe we have to decide on pairs of expressions here:

a) active and inactive scrollbar(s)
b) enabled and disabled scrollbar(s)

a) horizontal and vertical scrollbar(s)
b) scrollbar(s) along the x-axis and y-axis

a) scrolling mechanism
b) scrolling device
c) scrolling interface
d) scrollbars

My choice is 1a), 2a) and 3d). I just think that your
"mother/husband/roommate/brother/bus driver" would understand these
better. I personally do not know of any other visible scrolling
mechanism in UAs besides scrollbars.

> I'm not sure if I answered all your questions, let me know if I missed
> anything.
> ~fantasai

You have answered my questions. The only remaining issue that "bugs" me
a bit is this scrolling "panner" issue or UAs not using scrollbars but
some other *visible* scrolling mechanism. It's a minor issue; I am not
going to lose sleep over this.

regards, Gérard Talbot
Received on Monday, 21 December 2009 19:29:28 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 20 September 2010 17:52:01 GMT