On Mon, 10 Aug 2020 10:26:50 GMT, Nir Lisker <nlis...@openjdk.org> wrote:
>> In reading the existing doc carefully, it is both wrong and confusing. It >> has the wrong ordering of transform, scale >> and rotate. It is also confusing given that no distinction is made between >> matrix multiplication order and the observed >> effect on the geometry of the node. You have fixed the part about it being >> wrong, but the confusion remains, especially >> since you made a point of saying that the transforms list is applied in the >> reverse order. The matrices are multiplied >> in the following order: 1. layoutX/Y + translateX/Y/Z >> 2. rotate >> 3. scale >> 4. transforms[0] >> 5. transforms[1] >> ... >> >> We need to both list the actual matrix multiplication order, and then >> describe that the object is transformed from >> object coordinates to local coordinates of the node, to parent coordinated, >> etc., by first applying the transforms in >> the list in reverse order, then scale, rotate, translate+layout. I don't >> think that the `transforms` list is the place >> to do that (it belongs in the Transformations section). This doesn't seem >> like something we can sort out for JavaFX >> 15, so I would split this out and defer it. > > So should I revert all the changes about the transforms and leave it for 16, > or do part of it now and part later? > The changes are in the docs for the `Node` class, the `getTransforms()` > method and the `boundsInParentProperty()` > method. It seems best to revert all of the transform changes in this PR and deal with it in 16. ------------- PR: https://git.openjdk.java.net/jfx/pull/276