Closed Bug 441034 Opened 16 years ago Closed 16 years ago

Unfocused <select> arrow-button has transparent background

Categories

(Core :: Layout: Form Controls, defect)

x86
Windows Vista
defect
Not set
minor

Tracking

()

RESOLVED FIXED
mozilla1.9.1a1

People

(Reporter: nir123, Assigned: kliu)

References

Details

(Keywords: access, regression, testcase)

Attachments

(4 files)

Attached file testcase
Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1a1pre) Gecko/2008062103 Minefield/3.1a1pre

On unfocused <select> control, the arrow-button has transparent background.
The color of the arrow is black. Therefore, on <select> with black background, the arrow-button will be invisible. 

The bug occurs on Windows Vista theme only. Switching to "Windows Classic" will make the arrow-button visible.
Attached image screenshot
Severity: normal → minor
I've talked with Nir, and he know about some real-world websites which can demonstrate this issue. Basically, every website with a black background theme and a <select> on the page is affected, for example this online forum (please look at the bottom of the page): http://forums.hacking.org.il/viewforum.php?f=1
We should fall back to the old dropmarker style if the control has author-specified styling.  Not only is there the issue of transparency, but the new dropmarker style does not work well with anything other than native borders, and native borders are dropped if there is author styling.

Drop-down lists in XUL are currently rendered in a button-like style in Vista, so this is not applicable there.

I wonder if bug 434435 is a dupe of this; the reporter of that bug never posted a screenshot, so it's impossible to tell.
Assignee: nobody → kliu
Status: NEW → ASSIGNED
Attachment #326139 - Flags: superreview?(roc)
Attachment #326139 - Flags: review?(roc)
Attached image after patch
Attachment #326139 - Flags: superreview?(roc)
Attachment #326139 - Flags: review?(vladimir)
Attachment #326139 - Flags: review?(roc)
Keywords: checkin-needed
http://hg.mozilla.org/index.cgi/mozilla-central/rev/fe1683a02b0c
Status: ASSIGNED → RESOLVED
Closed: 16 years ago
Keywords: checkin-needed
Resolution: --- → FIXED
Target Milestone: --- → mozilla1.9.1a1
Comment on attachment 326139 [details] [diff] [review]
fall back to old style if there's author styling

Requesting 1.9.0.2 approval: very low-risk patch to fix an accessibility problem.
Attachment #326139 - Flags: approval1.9.0.2?
Can we get an automated test for this before landing on 1.9?
Flags: in-testsuite?
(In reply to comment #8)
> Can we get an automated test for this before landing on 1.9?
> 

I don't see how that could be done.  This is really a Vista theme issue, and as with all theme issues, there really isn't a way to test for this.  A check-if-equal refest won't work here, nor would a check-if-not-equal reftest (since the CSS needed to trigger the behavior would produce a difference in rendering anyway).
Couldn't you assert that applying that author style is not the same as applying that style as UA style, but _is_ the same as applying it as UA style and using "-moz-appearance: none"?
(In reply to comment #10)
> Couldn't you assert that applying that author style is not the same as applying
> that style as UA style, but _is_ the same as applying it as UA style and using
> "-moz-appearance: none"?
> 

That's a good idea.  But are we able to load agent and user sheets from a reftest?  I don't think that the stylesheet service is available to reftests, is it?
As I recall, you can now run chrome reftests, no?
Comment on attachment 326139 [details] [diff] [review]
fall back to old style if there's author styling

Pushing approval out to 1.9.0.3.
Attachment #326139 - Flags: approval1.9.0.2? → approval1.9.0.3?
Depends on: 450792
Attachment #326139 - Flags: approval1.9.0.4? → approval1.9.0.4-
Comment on attachment 326139 [details] [diff] [review]
fall back to old style if there's author styling

We're opting not to take this in a maintenance release.

But it should still get tests landed...
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: