Bug 209220 – [LegacySVG] SVG2: getScreenCTM(): ancestor CSS transform scale not applied

• Status: NEW
• Component: SVG

[Martin]

Created attachment 393829 [details]
run demo on ios device and check console output

Steps to reproduce the problem:
1. Visit https://jsfiddle.net/Lxj0cuso/2 or run attached demo with open console

What is the expected behavior?
Expected output:
ctm.a= 2
ctm.d= 2
scale: 2

What went wrong?
ios output:
ctm.a= 1
ctm.d= 1
scale: 2

On windows running chrome the a and d value of the matrix represent the scale factor (2) 
On iOS the values are both 1 but the scale is applied correctly,

Chrome version: 76.0.3809.81

[Alexey Proskuryakov]

I'm getting the reportedly incorrect output from Safari on macOS too.

[Radar WebKit Bug Importer]

<rdar://problem/60611550>

Created attachment 402280 [details]
Expected result (Chrome)

https://jsfiddle.net/z8y6bcrg/3/

I think I've run into the same issue but with a rotation on the ancestor.

I may be wrong about this, but given that both getScreenCTM() and getBoundingClientRect() fail under one condition or another (https://bugs.webkit.org/show_bug.cgi?id=212640), there doesn't appear to be any way to reliably translate a coordinate from a responsive SVG to client space...

This still appears broken in iOS 15.5. Have there been any updates on rdar://problem/60611550 which can be shared here?

[Ahmad Saleem]

*** Safari 16 on macOS 12.6 ***

ctm.a= 1
ctm.d= 1
scale: 2

*** Safari Technology Preview 154 ***

ctm.a= 1
ctm.d= 1
scale: 2

*** Firefox Nightly 107 ***

ctm.a= 1
ctm.d= 1
scale: 2

*** Chrome Canary 108 ***

ctm.a= 2
ctm.d= 2
scale: 2

____

Just wanted to share updated results. Thanks!

Is it something LBSE might fix?

[Nikolas Zimmermann]

Indeed fully fixed in LBSE - but not upstreamed yet. I also have new test cases for various scenarios. Stay tuned