620468 - "ASSERTION: viewport width must be nonnegative"

Status: RESOLVED FIXED

(Core :: SVG, defect)

Description

[Jesse Ruderman]

Attachment: testcase

###!!! ASSERTION: viewport width must be nonnegative!: 'aViewportWidth >= 0', file layout/svg/base/src/nsSVGUtils.cpp, line 797

Comment 1

[Jesse Ruderman]

Attachment: stack trace

Comment 2

[Daniel Holbert [:dholbert]]

This is a new assertion that I added in bug 272288 yesterday, to tighten up assumptions in nsSVGUtils::GetViewBoxTransform.

Negative widths aren't valid for <image> or <svg>, according to the spec:
 "A negative value is an error (see Error processing)"
 http://www.w3.org/TR/SVG11/struct.html#ImageElementWidthAttribute
 http://www.w3.org/TR/SVG11/struct.html#SVGElementWidthAttribute

Not sure if we currently trigger error-handling-code for this case...  If we don't want to bail out earlier than this, we could always just pass "max(0, width)" instead of "width" into nsSVGUtils::GetViewBoxTransform for cases like this.

Comment 3

[Robert Longson [:longsonr]]

We don't trigger error handling code as we don't distinguish between different types of length when we parse them. i.e. width (no negatives allowed) and x (negatives OK).

We just want to bail as early as possible in a way that makes mRect in nsSVGPathGeometryFrame the empty rect.

Comment 4

[Robert Longson [:longsonr]]

i.e. calling  nsSVGUtils::GetViewBoxTransform is pretty pointless if width <= 0 as we're not going to draw anything.

Comment 5

[Cameron McCormack (:heycam)]

Attachment: Bail out earlier when computing CTM of an <svg> element with negative width/height attributes.

Comment 6

[Cameron McCormack (:heycam)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/94423051f0fd

Comment 7

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/94423051f0fd