W3C

– DRAFT –
(MEETING TITLE)

11 June 2026

Attendees

Present
caribou, dmangal, karlcow, krit, nzimmermann
Regrets
-
Chair
Dirk Schulze
Scribe
krit

Meeting minutes

Clarify that getPathData() returns the attribute base value, not the CSS resolved/animated value

w3c/svgwg#1130

Vinay: That is about hte pathData implementation which is in FF and is about to get implemented in Chrome

viralipurbey:

viralipurbey: It is not clear which data should get returned during animation. baseVal or value of the path?

viralipurbey: SMIL and CSS Animations (any animations)

viralipurbey: Recommendation: getPathData() returns the base value of the d attribute for any animation.

viralipurbey: Firefox does this already

krit: Is there any other way to get the animation value in any other way? As part of WebAnimation for instance?

viralipurbey: yes, it is changing during animation.

krit: I mean any animation value

nzimmermann: animation like with WebAnimation APIs?

krit: yes

viralipurbey: There are only 3 end points and pathData just returns the base value.

viralipurbey: So there is no JS endpoint to get the animate value

nzimmermann: In WebAnimations, the getComputedStyle is reflecting the current state of the animation. So computed stlye should always reflect the current animation state.

<karlcow> https://svgwg.org/specs/paths/#InterfaceSVGPathData

nzimmermann: I need to verify it still though.

<karlcow> Hmm not the right url

viralipurbey: You are right, getComputedStyle returns the animated value.

<karlcow> https://w3c.github.io/svgwg/specs/paths/#InterfaceSVGPathData

krit: as long as there is some way to get the animated value I am fine

nzimmermann: this is also consistent with pathLength and other values.

RESOLUTION: getPathData() will return base value for any kind of animation

getPathSegmentAtLength(): Define which segment is returned at exact boundary

w3c/svgwg#1129

viralipurbey: this is another one but pathSegmentAtLength

viralipurbey: what happens if the length is on the boundary between 2 path segments.

nzimmermann: how would that be defined in the spec?

nzimmermann: how would we do it mathematically

krit: I'd say that we use SHOULD to avoid this discussion. It would be up to the implementation how to round the values and choose which segement to return

nzimmermann: fine with using the latter segment

RESOLUTION: Implementations SHOULD return the latter segment on boundaries of 2 segments. It is up to the implementation to define rounding rules/epsilon

ACTION: viralipurbey to update the spec

getPathSegmentAtLength(): Define behavior for NaN distance

w3c/svgwg#1128

viralipurbey: When we specify a distance it should be clammed between 0 and distance of path.

viralipurbey: What happens with NaN or Infinity in the distance?

viralipurbey: these are valid values, we just need to define the behavior

krit: FF returns NULL for NaN and Infinity the last sgement and -Infinity the first segment, what do other implementations?

nzimmermann: sounds mindful to me

nzimmermann: I'd like to avoid exceptions.

krit: wasn't getPathSegmentAtLength in Chrome?

viralipurbey: We deprecated the API and are now implementing the new API.

nzimmermann: you were thinking of getPathSegAtLength krit

nzimmermann: but it shows the point of viralipurbey : we were not throwing an exception for the old API

RESOLUTION: Follow behavior of Firefox: return NULL for NaN, Infinity the last sgment and -Infinity the first segment

Clarify glyph-orientation-vertical - whether it is obsolete or something need to be kept?

w3c/svgwg#1122

karlcow: I was working on it and was writing tests when Robert mentioned it is not in Chrome nor Firefox. Should we keep it or drop it?

karlcow: One section says it has been marked obsolet and the next section that it has been replaced.

karlcow: The spec text seems to be confusing though.

nzimmermann: I would support it. We only implemented because one test in the test suite.

nzimmermann: The text also explicitly did exclude horizontal.

nzimmermann: However, for vertical text layouts in a stand alone context it could be useful.

nzimmermann: you do not want to deal with individual glyphs. But if you want to do layout as usual but rotate the individual glyphs. It could be implemented in InkScape.

<dmangal> https://wpt.fyi/results/css/css-variables/variable-presentation-attribute.html?label=experimental&label=master&aligned

Vinay: It seems not to be supported by most browser engines and Webkit agrees to remove it, so shall we remove it?

ACTION: Karl to contact Tav on the implementation status

<Image> auto-rotation due to 'Orientation' attribute in EXIF Data

w3c/svgwg#819

nzimmermann: It is about an attribute that can be embedded in the EXIF DATA

nzimmermann: there is no attribute in SVG to control if the data point would be used in SVG or not.

nzimmermann: there is a request to implement it. Check the last comment in the ticket.

nzimmermann: 3 options: To repeat, I see at least 3 options:

(a) Always apply image orientation EXIF data in

(b) Use the CSS image-orientation property from the element with the filter property for that instance of the filter.

(c) Define an attribute on to control the orientation (analogous to the CSS property).

nzimmermann: I would need to get more context on some options

nzimmermann: We have the image element which is a defined filter resource.

nzimmermann: If you apply the filter some where, we need a way to choose to honor the data on the source where the filter gets applied

krit: lost here: what is the relation to filter?

nzimmermann: we need to think about more than image like the feImage element

viralipurbey: is it clear for image element or do we still need to figure it out

nzimmermann: we need to figure it out in all situations. This is a request from CSS WG: Do we want to expose it via attribute?

nzimmermann: I missed that this is beyond image initially so I need to think about the other use case more.

nzimmermann: so do we always want to auto-orientate which is a privacy concern by chrome? Or do we want to expose it explictly?

karlcow: there is a similar discussion on HTML image

karlcow: the issue is closed but I don't know the conclusion right now

<karlcow> whatwg/html#4495

<nzimmermann> https://developer.mozilla.org/en-US/docs/Web/CSS/Reference/Properties/image-orientation

krit: if it is a privacy concern, what does it matter if it is auto or explicitly?

nzimmermann: I am confused by this myself.

<karlcow> https://html.spec.whatwg.org/multipage/imagebitmap-and-animations.html#imageorientation

nzimmermann: we would need more clarification: What are the privacy concerns? What does it matter if auto or explict?

viralipurbey: it is about cross origin leakage

viralipurbey: EXIF metadata could be leaked

krit: What does HTML do? Aren't the concerns valid there too then?

nzimmermann: I agree: Do not invent our own thing but do what HTML does and unerstand why feImage is special.

nzimmermann: that part might be related to the privacy concern.

ACTION: On Niko to get more information from Stephen about it.

TPAC

caribou: We need to create a github issue to request a room

karlcow: we should use the momentum and meet

<caribou> https://lists.w3.org/Archives/Member/chairs/2026AprJun/0067.html

<caribou> w3c/tpac2026-meetings

<caribou> w3c/strategy#559

recharter

caribou: We created a git ticket for recharter. Currently we have maintenance as main topic. If you have more ideas, let me know

karlcow: I don't know if we can do much more than finishing SVG2 for the next period

caribou: questions will likely be more about resources and not topics. Discussion will happen on github

caribou: announcement from/to AC will be send soon. Probably next week.

Summary of action items

  1. viralipurbey to update the spec
  2. Karl to contact Tav on the implementation status
  3. On Niko to get more information from Stephen about it.

Summary of resolutions

  1. getPathData() will return base value for any kind of animation
  2. Implementations SHOULD return the latter segment on boundaries of 2 segments. It is up to the implementation to define rounding rules/epsilon
  3. Follow behavior of Firefox: return NULL for NaN, Infinity the last sgment and -Infinity the first segment
Minutes manually created (not a transcript), formatted by scribe.perl version 248 (Mon Oct 27 20:04:16 2025 UTC).

Diagnostics

Succeeded: s/0/NaN/

Maybe present: Vinay, viralipurbey

All speakers: caribou, karlcow, krit, nzimmermann, Vinay, viralipurbey

Active on IRC: caribou, dmangal, karlcow, krit, nzimmermann