199897 - Some areas are not repainted when scrolling (misplaced pixel)

Status: RESOLVED FIXED

(Core :: Web Painting, defect, P2)

Description

[rbs]

Some areas are not repainted when smoothscroll is enabled. This is probably due
to roundoff errors. I tried the patch on attachment 118792 [details] [diff] [review] from bug 199159, but
it didn't make a difference. I will attach a screen capture to show the problem.

Comment 1

[rbs]

Attachment: screnshoot - see the trailing dots below the <hr>

sorry, a .bmp image -- I don't have a tool at hand right now for the
conversion. It is viewable with Mozilla/Nav7.

Comment 2

[Sébastien Deleuze]

I have Mozilla 1.4a, and this bug exist also when smooth scrolling is disabled,
as you can watch it at the left of this page :
http://linuxfr.org/2003/04/02/11938.html

Comment 3

[OlivierAJ]

I've got the same problem on Windows NT (SP4) and Mozilla 1.4a official release
(Gecko/20030401). This problem appear with "general.smoothScroll" True OR False

Comment 4

[Stefan Borggraefe]

I see this bug with Linux and a fresh CVS-Build.

This didn't happen a few weeks ago, so this is a regression.

Comment 5

[Stefan Borggraefe]

*** Bug 199551 has been marked as a duplicate of this bug. ***

Comment 6

[Stefan Borggraefe]

This has nothing to do with smoothscroll (see comment #3 and bug 199551. Also I
tested it myself). The misplaced pixel appear when you scroll the page. With
smooth scroll there are more of them because more (finer-grained) scroll steps
are performed.
Adding URL from comment #2 to the URL field of the bug.

Comment 7

[Robert O'Callahan (:roc) (email my personal email if necessary)]

This was a regression caused by the fix for bug 86249.

The problem is that when the dirty rect is entirely outside the border box,
border clipping creates rectangles with negative widths and/or heights and then
strange things happen. The fix is to not clip if we would create a bogus rectangle.

Comment 8

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: fix

I checked that this preserves the fix for the testcases in bug 86249.

Comment 9

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Why can't you do an Intersects check earlier in the function and just bail if
they don't intersect?  (Should PaintBackground do the same thing?  Or should the
caller be responsible for that?)

Comment 10

[Robert O'Callahan (:roc) (email my personal email if necessary)]

> Why can't you do an Intersects check earlier in the function and just bail if
> they don't intersect?

If dirtyRect intersects the outerRect but not the innerRect, we have to draw
some border and can't bail. But we still don't want the innerRect to get a bad size.

> Should PaintBackground do the same thing?  Or should the
> caller be responsible for that?

The caller usually does do this via nsContainerFrame::Paint (I think). HR frames
are special because the border area is smaller than the frame rect.

Comment 11

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: better patch

this patch is safer and clearer, I think.

Comment 12

[Stefan Borggraefe]

Small rendering errors on many pages make Mozilla look bad and there is a fix.
Maybe this should block Mozilla 1.4b.

Comment 13

[Michael Lefevre]

*** Bug 204552 has been marked as a duplicate of this bug. ***

Comment 14

[Robert O'Callahan (:roc) (email my personal email if necessary)]

*** Bug 205227 has been marked as a duplicate of this bug. ***

Comment 15

[Robert O'Callahan (:roc) (email my personal email if necessary)]

The risk here is very small, and I think the gain is significant ... leaving
painting dribbles on pages with HRs just looks shoddy.

Comment 16

[Stefan Borggraefe]

I have tested the patch for two weeks now on Linux. It fixes the bug and I
encountered no new problems.

Comment 17

[Asa Dotzler [:asa]]

Comment on attachment 121610 [details] [diff] [review]
better patch

a=asa (on behalf of drivers) for checkin to 1.4b.

Comment 18

[Asa Dotzler [:asa]]

oops, 1.4, not 1.4b

Comment 19

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Fix checked in.

Comment 20

[José Jeria]

*** Bug 206154 has been marked as a duplicate of this bug. ***