9285 - background image overflow on first load

Status: VERIFIED FIXED

(Core :: Layout, defect, P3)

Description

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

The "worms" background image in the above test case overflows outside of the
first paragraph that has it (but not the small one later) the *first* time the
page is loaded off the network.  It seems the image overflows up to the image
boundary.  It should, instead, be clipped, as it is on any later reflow or
reload.

Comment 1

[rickg]

This appears to be another background image bug. Please take a look.

Comment 2

[dcone (gone)]

Fixed

Comment 3

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

In 1999-07-17-08-M8 (Linux), this is worse than it was before.  It now doesn't
go away on later reflows, and it overflows out of the second paragraph as well.

Reopening bug.

Comment 4

[dcone (gone)]

This is fixed as of the 7/19/99 build.  The bug report and the fixed were not
sync'ed.  Ends up being a clipping bug in the background render'er

Comment 5

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

It still flashes the incorrect display before showing the correct one.  I'm not
going to verify it yet.  That means (to me, anyway):
* it's not completely fixed
* there's an extra repaint/reflow somewhere

Comment 6

[Chris Petersen]

With the latest Linux build (July 26th 1999072614) , the page is rendered
correctly.

Comment 7

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

Reopening bug.

In apprunner linux 1999-08-13-08-M9, I'm still seeing the overflow *while* the
document is loading (perhaps before the second image loads?), but then it seems
to refresh and it goes away.  I only see the problem on the first load.

Comment 8

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

I'm seeing similar problems during progressive loading of

http://www.fas.harvard.edu/~dbaron/css/test/sec110102

Comment 9

[dcone (gone)]

This seems to be some sort of reflow issue.  The backgrounds are painted
correctly, and only when all the images are loaded.  Until they are loaded there
may be a reflow that does not all the URL's, images, etc.  Anyway this is not a
background problem, it may be a reflow problem, so I am giving it to Patrick to
check out and disposition from here.

Comment 10

[/\/\arcio Galli]

Adding testcases related with background of DIV area which have an image:

Image not absolutely positioned inside DIV:
http://www.taboca.com/jan00nscp/dom/demos/dom2/on/index35.html
Image is small but background is size of screen minus border - just like table 
with width=100%

Image with absolute positioning:
http://www.taboca.com/jan00nscp/dom/demos/dom2/on/index33.html
bg goes to infinity to the right 

Found at this demo:
http://www.taboca.com/jan00nscp/dom/demos/dom2/on/index3.html
(try dragging this DIV area). See that entire area infinitely to right is 
receiving onmousedown. 

Confirmated with 2000022408, win98

Comment 11

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

I'm seeing similar (except the reverse) problems the first time
http://www.w3.org/Graphics/PNG/inline-alpha.html
is loaded:  the background image behind the first test is not completely tiled
on the first load.  (The repaint caused by the loading of the foreground image
fixes some of the problem.)  Screenshot to be attached.

Comment 12

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

Attachment: screenshot of problem on inline-alpha.html

Comment 13

[Patrick C. Beard]

Giving to correct module owner.

Comment 14

[Kevin McCluskey (gone)]

Reassigning to dcone- Triaging Troy's bug list

Comment 15

[dcone (gone)]

This is that invalidate problem.. I think you fixed.. but I am not sure.

Comment 16

[Marc Attinasi]

This looks like it was fixed when I fixed bug 36811 (incorrect invalidation 
of image bounds instead of frame bounds when the image load is completed). I 
looked at several of the testcases and they look correct. The original URL is 
certainly better - the worms are now bounded by their paragraph.

Marking Fixed.

Comment 17

[Chris Petersen]

With the Linux May 22nd build, I'm not reproducing this problem. Marking verified 
fixed.