276414 2024-07-10 02:26:48 -0700 Handling of negative margins in MathML 2025-07-07 05:53:44 -0700 1 1 1 Unclassified WebKit MathML WebKit Nightly Build Unspecified Unspecified RESOLVED WORKSFORME InRadar P2 Normal --- 276428 276169 1 fred.wang fred.wang webkit-bug-importer oldest_to_newest 2045499 0 fred.wang 2024-07-10 02:26:48 -0700 It seems margin has been supported for mrow-like elements for a while and I've been doing that for other MathML elements in bug 276169's dependencies. MathML elements calculate their sizes from the coordinates/sizes of their children's margin boxes, which can lead to weird cases with negative margins. We should make sure these edge cases don't cause any wrong behavior laying MathML elements, probably clamping negative offsets to 0 when necessary. 2045540 1 fred.wang 2024-07-10 08:05:51 -0700 There are WPT crash tests for what I'm mentioning: http://wpt.live/mathml/crashtests/children-with-negative-inline-sizes.html http://wpt.live/mathml/crashtests/children-with-negative-block-sizes.html and they don't seem to cause any crash/asserts, so maybe some explicit clamping is not necessary. 2046685 2 webkit-bug-importer 2024-07-17 02:27:15 -0700 <rdar://problem/131909042> 2128122 3 fred.wang 2025-07-07 05:53:44 -0700 I'll close this issue. I believe MathML Core does not support negative margin for now and WPT tests mentioned above don't crash/assert. If things change we can always revisit.