Allow non-borders to be adjusted to less than 1 when zoomed out

• Component: Layout and Rendering
• Status: RESOLVED FIXED

[Emil A Eklund]

Due to how margins are handled for floats the width might be off by one when zoomed out. This causes problems for certain sites that use floats and hard coded widths. Most notably the New York Times website, www.nyt.com

Downstream chromium bug: http://code.google.com/p/chromium/issues/detail?id=133942

[Emil A Eklund]

The problem seems to be that the margin never shrinks below 1 causing the width including the margin to extend past the container for zoom levels below 100%.

[Emil A Eklund]

(In reply to comment #1)
> The problem seems to be that the margin never shrinks below 1 causing the width including the margin to extend past the container for zoom levels below 100%.

...and that logic is in CSSPrimitiveValue::computeLengthDouble:

    // Any original result that was >= 1 should not be allowed to fall below 1.  This keeps border lines from
    // vanishing.
    double zoomedResult = result * multiplier;
    if (zoomedResult < 1.0 && result >= 1.0)
        return 1.0;

This should probably only kick in for borders (which are represented as integers).

[Emil A Eklund]

Created attachment 150003 [details]
Patch

[Emil A Eklund]

CC:ing hyatt who added the check in http://trac.webkit.org/changeset/31219 back in the dark ages of 2008.

[Eric Seidel (no email)]

Comment on attachment 150003 [details]
Patch

Makes sense to me.

[WebKit Review Bot]

Comment on attachment 150003 [details]
Patch

Rejecting attachment 150003 [details] from commit-queue.

New failing tests:
svg/zoom/page/zoom-img-preserveAspectRatio-support-1.html
fast/transforms/bounding-rect-zoom.html
svg/zoom/page/zoom-replaced-intrinsic-ratio-001.htm
fast/reflections/reflection-with-zoom.html
Full output: http://queues.webkit.org/results/13109213

[WebKit Review Bot]

Created attachment 150046 [details]
Archive of layout-test-results from ec2-cq-02

The attached test failures were seen while running run-webkit-tests on the commit-queue.
Bot: ec2-cq-02  Port: <class 'webkitpy.common.config.ports.ChromiumXVFBPort'>  Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick

[WebKit Review Bot]

Comment on attachment 150003 [details]
Patch

Attachment 150003 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/13107333

New failing tests:
svg/zoom/page/zoom-img-preserveAspectRatio-support-1.html
fast/transforms/bounding-rect-zoom.html
svg/zoom/page/zoom-replaced-intrinsic-ratio-001.htm
fast/reflections/reflection-with-zoom.html

[WebKit Review Bot]

Created attachment 150088 [details]
Archive of layout-test-results from ec2-cr-linux-01

The attached test failures were seen while running run-webkit-tests on the chromium-ews.
Bot: ec2-cr-linux-01  Port: <class 'webkitpy.common.config.ports.ChromiumXVFBPort'>  Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick

[Emil A Eklund]

Created attachment 150269 [details]
Patch for landing

[Emil A Eklund]

Committed r121599: <http://trac.webkit.org/changeset/121599>

[Luke Macpherson]

Comment on attachment 150269 [details]
Patch for landing

View in context: https://bugs.webkit.org/attachment.cgi?id=150269&action=review

> Source/WebCore/css/StyleBuilder.cpp:611
> +                length = zoomedLength;

It's worth noting that ApplyPropertyComputeLength is used by lots of CSS properties, not just borders, so you may want to reduce the number of cases where this is applied even further.