300125 - Carets from other tabs appear on the same tab when re-enabling caret browsing mode

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[Uri Bernstein]

When multiple tabs are open in caret browsing mode, tuning caret browsing off
and then on again causes multiple cartes to appear on the same tab. The "ghost"
carets are blinking, and do not move as the page is scrolled.

Steps to reproduce:
0. Caret mode is off
1. Open two tabs, with a HTML page in each of them
2. Enable caret browsing mode (F7)
3. On each of the tabs, place the caret somewhere on the page
4. Disable caret browsing mode
5. Enable caret browsing mode

You will see that both carets appear, and thy are both blinking.
The same could be done with more than two tabs, resulting in more than two
blinking carets.

This seems to be an old bug (I can see it in Firefox 1.0), but I couldn't find a
duplicate.
Not sure if this is Mac-specific (can someone please try on other platforms?)

Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8b3) Gecko/20050707
Firefox/1.0+

Comment 1

[Uri Bernstein]

Oops, this also affects SeaMonkey, so back to Core.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Sounds like an issue with how caret browsing works (or possibly with caret itself).  Not a tabs issue.

Comment 3

[Ginn Chen]

reproduced on Fx 1.0.7, Deer Park, 1.5RC on Mac OS X 10.4.3
can not reproduced on Linux

I don't know think it related to the caret disappearance bug whiling switching tabs, since that patch is not integrated into Fx 1.0.7.

Comment 4

[Ginn Chen]

I added a printf in nsRenderingContextGTK::InvertRect and nsRenderingContextMac::InvertRect.
I got the same result.
That means when I follow the steps in description, GTK build also blinking 2 carets.
But on Linux, carets on different tabs use different drawables, one of them is drew behind.
After you switch tabs, only one caret is blinking.

According to this, I think this bug should be fixed in layout/ or content/event/ rather than gfx/src/mac.

Comment 5

[Ginn Chen]

Attachment: patch

I didn't find a regression yet.

Don't get confused with bug 306173 when testing it.

Comment 6

[Johnny Stenback (:jst)]

Comment on attachment 202345 [details] [diff] [review]
patch

sr=jst

Comment 7

[Ginn Chen]

Checking in nsEventStateManager.cpp;
/cvsroot/mozilla/content/events/src/nsEventStateManager.cpp,v  <--  nsEventStateManager.cpp
new revision: 1.625; previous revision: 1.624
done