dbaron is concerned that we set the NS_FRAME_DESCENDANT_INTRINSIC_ISIZE_DEPENDS_ON_BSIZE flag in cases where we don't really need to. Quoting him from bug 1242172 comment 14: === It seems to me that we set NS_FRAME_DESCENDANT_INTRINSIC_ISIZE_DEPENDS_ON_BSIZE more aggressively than needed in that we don't check that the specified block-size that we're doing it for is a percentage block-size rather than a length. It also seems to me that we propagate NS_FRAME_DESCENDANT_INTRINSIC_ISIZE_DEPENDS_ON_BSIZE up the tree more aggressively than needed, since we ought to be able to stop propagating after reaching an element whose block-size does not depend on the block-size of its containing block (although after marking that one). ===