2564 - [DOGFOOD] [PP] Scrolling a long document yields text overwriting itself

Status: VERIFIED FIXED

(Core :: Layout: Form Controls, defect, P1)

Description

[Simon Fraser [no longer active]]

Load the above URL in viewer, and wait for it to finish loading. Then scroll
down. Notice that there are large areas which are completely blank (I see
two of them), as if some big chunks of the document have been rendered at
the wrong offset. There is nothing in the HTML that should cause these blank
areas.

This does not occur on Windows, where the entire file has no gaps.

An additional test would be to scroll the document up and down while it is
rendering, and see if that causes any positioning problems.

Comment 1

[Pierre Saslawsky]

Bug #2385 also refers to big blank sections on the Mac.

Comment 2

[Pierre Saslawsky]

Bug #2385 has been closed as dup of #2615. Apparently this problem and the one in
#2385/#2615 are different.

Comment 3

[leger]

Inserting Milestone info.

Comment 4

[leger]

Setting all current Open/Normal to M4.

Comment 5

[Paul MacQuiddy]

per leger, assigning QA contacts to all open bugs without QA contacts according
to list at http://bugzilla.mozilla.org/describecomponents.cgi?product=Browser

Comment 6

[Pierre Saslawsky]

Changed target to M5

Comment 7

[Pierre Saslawsky]

I tried with a recent build on Windows and I got a very similar problem except
that instead of "big blank sections", we have big "black" sections. The "black"
sections contain text that is overlayed over and over.

Reassigned to kipp.

Comment 8

[Pierre Saslawsky]

Bug #1117 might be related to this one.

Comment 9

[Patrick C. Beard]

I've fixed this on the Mac, still puzzled on Win95.

Comment 10

[kipp]

On linux we get smearing - it seems that the drawing of the background fails in
these situations. I suspect that the coordinate math isn't handling the large
coordinates that this results in (3,465,066 twips or 231,004 pixels!)

Comment 11

[Kevin McCluskey (gone)]

This is probably a coordinate overflow problem. On Win95
nsRenderingContext::SetClipRect and nsRenderingContext::FillRect need to have
the rectangles that they pass to the native rountines conditioned to prevent the
top and bottom of the RECT to be within WIN95's acceptable range. This is
probably what GTK needs as well.

Comment 12

[leger]

Moving all Widget Set bugs, past and present, to new HTML Form Controls
component per request from karnaze.  Widget Set component will be retired
shortly.

Comment 13

[ramiro]

m10

Comment 14

[ramiro]

Attachment: sample long doc

Comment 15

[ramiro]

see also:

http://bugzilla.mozilla.org/showattachment.cgi?attach_id=1357

Comment 16

[ramiro]

Changed URL, marking assigned, m11.

I see the bug, the problem is that the arguments to XFillRectangle()
(gdk_draw_rectangle() actually) are overflowing at -(2^15) and +(2^16) cause of
the 'unsigned int' limit of x windows.

So, im going to add some code to clamp all arugments to X lib functions.

Comment 17

[ramiro]

*** Bug 12298 has been marked as a duplicate of this bug. ***

Comment 18

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

This causes problems with http://www.w3.org/TR/REC-CSS1 once you've scrolled
about 2/3 of the way down.  I think it's the same problem as bug 1177 (not
1117!), which covered the fixes for Mac and Win95.

Comment 19

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

There are some comments relating to this bug on bug 6048.

Comment 20

[kipp]

*** Bug 13389 has been marked as a duplicate of this bug. ***

Comment 21

[leger]

*** Bug 13657 has been marked as a duplicate of this bug. ***

Comment 22

[Mats Palmgren (inactive)]

*** Bug 14212 has been marked as a duplicate of this bug. ***

Comment 23

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

*** Bug 13690 has been marked as a duplicate of this bug. ***

Comment 24

[ramiro]

I see the bug, the problem is that the arguments to XFillRectangle()
(gdk_draw_rectangle() actually) are overflowing at -(2^15) and +(2^16) cause of
the 'unsigned int' limit of x windows.

So, im going to add some code to clamp all arugments to X lib functions.

Comment 25

[Eli Goldberg]

*** Bug 16045 has been marked as a duplicate of this bug. ***

Comment 26

[Mats Palmgren (inactive)]

*** Bug 16174 has been marked as a duplicate of this bug. ***

Comment 27

[Eli Goldberg]

As a courtesy, I've changed the bug description to reflect the behavior reported

by the many people who are still reporting this bug, in the hopes that it may be

easier for them to find. (Both here, and on the Most Frequent Bugs page.)

Comment 28

[Simon Fraser [no longer active]]

Reassign to pavvy pav pav

Comment 29

[kipp]

*** Bug 17238 has been marked as a duplicate of this bug. ***

Comment 30

[Stuart Parmenter]

*** Bug 16809 has been marked as a duplicate of this bug. ***

Comment 31

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

Attachment: patch to fix bug

Comment 32

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

The above patch fixes the problem for me.  You might want to fiddle with the
numbers a little (perhaps even see how gdk_draw_rectangle works and give it more
margin for error if you think it's necessary).  Also, I'm not sure if it's
needed on InvertRect.  It probably is for DrawRect, though, since someone might
put a border around the CSS1 spec someday...

Comment 33

[leger]

Putting on PDT+ radar.

Comment 34

[Stuart Parmenter]

looks good.  checked in fix.

Comment 35

[Chris Petersen]

Fixed in the Nov 12 Linux build.