W3C

– DRAFT –
SVG WG

02 July 2026

Attendees

Present
karlcow, krit, nzimmermann, Ragvesh, ydaniv
Regrets
-
Chair
Dirk Schulze
Scribe
krit

Meeting minutes

Remove instanceRoot and animatedInstanceRoot

w3c/svgwg#852

<karlcow> I don't hear anything

karlcow: I looked at it last week: it is declared in the spec...

karlcow: but it is not implemented

krit: is that new to SVG2?

nzimmermann: it is SVG1.1 and used to be mandatory and important and pre-dates shadow root

nzimmermann: when you reference an element with use, browsers needed to build an instance tree. Events were part of the instance. This was dropped with SVG 2 and ShadowDom

nzimmermann: so this API is now useless and agree to drop it

RESOLUTION: Remove instanceRoot and animatedInstanceRoot in favor for ShadowDOM

Media fragment percent fallback should reference css-images-3 default object size instead of CSS 2.1

w3c/svgwg#1136

<dmangal> dmangal

dmangal: We discussed it before

dmangal: when media fragments are involved and there is no witdth and height on SVG root, we agreed to follow the user style sheet with 500x250

dmangal: but what should happen if a viewbox is present but not width/height? What if only one is missing?

<karlcow> https://www.w3.org/TR/css-images-3/#default-sizing

dmangal: I suggest to use the default sizing algorithm in CSS3

dmangal: Firefox already follows the new spec text and Chrome is about to follow the newer default sizing algorithm

karlcow: I don't think it is an issue

karlcow: It doesn't pass in Safari right now but it would be expected. So agree to the proposal

nzimmermann: I wasn't aware of the sizing algorithm. I like it.

karlcow: the algorithm siziing has been implemented in WebKit but WPT doesn't pass.

dmangal: maybe because media fragments are not supported in WebKit.

karlcow: probably.

RESOLUTION: Media fragment percent fallback should reference css-images-3 default object size

ACTION: dmangal will update the spec text

SVGAngle.animVal.unitType during animation between different units is unspecified

w3c/svgwg#1146

viralipurbey: this is about the angle in SVG. When animating from one angle to another, and the unit types differ, the spec doesn't say what the unit type should be.

viralipurbey: Chrome and WebKit return a canonical unit

viralipurbey: but there is a WPT test failing in Safati and Chrome but passing in Firefox. The spec is not clear which implementation is right

nzimmermann: I'd rather go to the baseVal unit. If the animation goes away, we would fallback to baseVal.

nzimmermann: this would be consistent and independent of from-to-by animation types if we stick to the base-val

viralipurbey: I'd like to add that CSS uses canonical units

nzimmermann: I wasn't aware. Maybe we should follow CSS?

nzimmermann: what about length and other types?

nzimmermann: like cm to inch?

nzimmermann: would the solution for SVGAngle apply to all types, not just angles?

viralipurbey: let me check that

<karlcow> https://wpt.fyi/results/svg/animations?label=master&label=experimental&aligned&q=unitType

nzimmermann: using the baseVal can be helpful for devs because they can specify it.

karlcow: SVGLength seems to be passing for everyone

karlcow: it is only testing pixel to cm

ACTION: karlcow viralipurbey to test behavior across all types and units to find a common ground

<dmangal> w3c/csswg-drafts#14028

<dmangal> w3c/csswg-drafts#14028

[circle][ellipse] SVG 2 silent on negative values for cx/cy

<karlcow> w3c/svgwg#1145

w3c/svgwg#1145

karlcow: The SVG2 spec does not say negative values are forbidden for geometry attributes (specifically rx and ry) only makes sense for x and y?

krit: what happens if r, rx, ry are negative in implementations today

karlcow: In Chrome+Firefox it goes to zero

viralipurbey: It is fair to have a value greater zero

<karlcow> WebKit/WebKit#68442

dmangal: we might print something if rx is valid but ry isn't

<karlcow> https://bug-318324-attachments.webkit.org/attachment.cgi?id=480272

krit: so you suggest ignore negative values.

karlcow: yes, Firefox is consistent. Blink and Webkit have a different behavior

karlcow: the fallback value is 0

karlcow: ... in Firefox it is 0

nzimmermann: that is the default, no?

krit: default is auto.

karlcow: suggest to go to 0

karlcow: baseVal stays negative for FF and Blink. But getComputedStyle is 0.

karlcow: for rectangle the getComputedStyle would be auto and FF is 0.

dmangal: The spec already mentioned that computed style is auto and resolves to 0

karlcow: the spec says it for r and rx and ry but not for dimension values

nzimmermann: so lets make it consistent for all values

karlcow: yes

viralipurbey: then we need to define it for all objects

viralipurbey: there should be a fallback

karlcow: suggesting to follow FF: computed style would be auto and resolved values to 0

dmangal: width and height it may resolve to 100% or 0

dmangal: lets say an SVG outer SVG element with negative width and height should be 100% and not 0

<karlcow> https://w3c.github.io/svgwg/svg2-draft/geometry.html#Sizing

krit: So you suggest for shape elements it would resolve to 0, but for outer SVG it could be 100%?

<karlcow> Svg element -> auto -> 100%

<karlcow> All other elements -> auto -> 0

RESOLUTION: negative length values on presentation attributes that do not allow negative values compute to auto and resolve to 0. Need to define the behavior on non geometry elements later.

<karlcow> Image -> auto -> 300x150

RRSAgent: make minutes

Summary of action items

  1. dmangal will update the spec text
  2. karlcow viralipurbey to test behavior across all types and units to find a common ground

Summary of resolutions

  1. Remove instanceRoot and animatedInstanceRoot in favor for ShadowDOM
  2. Media fragment percent fallback should reference css-images-3 default object size
  3. negative length values on presentation attributes that do not allow negative values compute to auto and resolve to 0. Need to define the behavior on non geometry elements later.
Minutes manually created (not a transcript), formatted by scribe.perl version 248 (Mon Oct 27 20:04:16 2025 UTC).

Diagnostics

Succeeded: s/specified value/getComputedStyle/

Maybe present: dmangal, RRSAgent, viralipurbey

All speakers: dmangal, karlcow, krit, nzimmermann, RRSAgent, viralipurbey

Active on IRC: dmangal, karlcow, krit, nzimmermann, Ragvesh, ydaniv