On Thu, 15 Oct 2026 12:06:37 GMT, Florian Kirmaier <[email protected]> wrote:
>> modules/javafx.graphics/src/main/java/javafx/scene/layout/GridPane.java line >> 2067: >> >>> 2065: final double change = Math.abs(limit) <= >>> Math.abs(portion)? limit : portion; >>> 2066: heights.setSize(index, isSnapToPixel() ? >>> ScaledMath.rint(heights.getSize(index) + change, getSnapScaleY(this)) : >>> heights.getSize(index) + change); >>> 2067: available = isSnapToPixel() ? >>> Math.rint((available - change) / pixelSize) * pixelSize : available - >>> change; >> >> `change` is not rounded, this might fail at scales != 100% > > as far as i see, change is already a multiple of pixelSize. > So i think it's correct. no, we do have a regression here: - scale 1.5 - no gaps - three ALWAYS-growing rows, each with a node pref=10 - one child node that spans all three rows, pref=38 - resize to height 44 result: the node that spans all three is one pixel short ------------- PR Review Comment: https://git.openjdk.org/jfx/pull/2312#discussion_r4286404471
