Bug 95551 – REGRESSION (r127140): fast/forms/datalist/input-appearance-range-with-datalist-rtl.html is failing

• Status: NEW
• Component: Tools / Tests

[Sudarsana Nagineni (babu)]

fast/forms/datalist/input-appearance-range-with-datalist-rtl.html is failing on EFL bots after r127140.

http://trac.webkit.org/changeset/127140

http://build.webkit.org/results/EFL%20Linux%2064-bit%20Debug/r127141%20%285152%29/fast/forms/datalist/input-appearance-range-with-datalist-rtl-diffs.html

[Raphael Kubo da Costa (:rakuco)]

The "cause" seems to be the reftest itself.

It works fine on, say, Chromium, because it does not seem to use a background filling color for the amount the slider has been dragged. Since the EFL port does that, expecting <input type=range value=0> to render the same as <input type=range value=100 dir=rtl> is not a valid assumption.

Keishi, do you have any ideas on how we can fix the expectations here?

[Keishi Hattori]

(In reply to comment #1)
> The "cause" seems to be the reftest itself.
> 
> It works fine on, say, Chromium, because it does not seem to use a background filling color for the amount the slider has been dragged. Since the EFL port does that, expecting <input type=range value=0> to render the same as <input type=range value=100 dir=rtl> is not a valid assumption.
> 
> Keishi, do you have any ideas on how we can fix the expectations here?

I didn't realize we had unsymmetrical sliders. I will change this test to a pixel test.

[Keishi Hattori]

Created attachment 166496 [details]
Patch

[Raphael Kubo da Costa (:rakuco)]

Any reason to require an empty text dump besides the pixel expetation?

[Keishi Hattori]

(In reply to comment #4)
> Any reason to require an empty text dump besides the pixel expetation?

No. How do I get rid of it?

[Raphael Kubo da Costa (:rakuco)]

(In reply to comment #5)
> (In reply to comment #4)
> > Any reason to require an empty text dump besides the pixel expetation?
> 
> No. How do I get rid of it?

I guess it was added because you called dumpAsText(true) in the test.

[Keishi Hattori]

(In reply to comment #6)
> I guess it was added because you called dumpAsText(true) in the test.

I used dumpAsText because without it I get a render tree dump. And a render tree dump will be different for each port.

[Keishi Hattori]

Created attachment 166503 [details]
Moved input-appearance-range-with-datalist-rtl-expected.txt

[Raphael Kubo da Costa (:rakuco)]

(In reply to comment #7)
> (In reply to comment #6)
> > I guess it was added because you called dumpAsText(true) in the test.
> 
> I used dumpAsText because without it I get a render tree dump.

Doh, right :-)

> And a render tree dump will be different for each port.

Is this really a problem? The PNG expectations already differ anyway.

[WebKit Review Bot]

Comment on attachment 166503 [details]
Moved input-appearance-range-with-datalist-rtl-expected.txt

Attachment 166503 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/14084901

New failing tests:
fast/forms/datalist/input-appearance-range-with-datalist-rtl.html

[Ryosuke Niwa]

Comment on attachment 166503 [details]
Moved input-appearance-range-with-datalist-rtl-expected.txt

View in context: https://bugs.webkit.org/attachment.cgi?id=166503&action=review

> LayoutTests/fast/forms/datalist/input-appearance-range-with-datalist-rtl.html:6
> +    testRunner.dumpAsText(true);

Why do we not want to have render tree dump? That makes this test useless on all but Chromium port.f

[Darin Adler]

Comment on attachment 166503 [details]
Moved input-appearance-range-with-datalist-rtl-expected.txt

Maybe instead we can make this a reference test by using LTR sliders that look the same because of reversed data as the reference.

[Anders Carlsson]

Comment on attachment 166503 [details]
Moved input-appearance-range-with-datalist-rtl-expected.txt

Clearing review flag on patches from before 2014. If this patch is still relevant, please reset the r? flag.