Glen, I've detected a regression in the handling of writing-mode. The two issues also existed in my original performance patch and is completely my fault. 1) When parwmrel2abs is used, it must be the parent-FO's writing-mode which is used. 2) The mapping of writing-mode to one of the 3 arguments in PropertyList.wmMap() contains a copy-paste error.
Created attachment 9858 [details] A unified diff against HEAD.
Created attachment 9859 [details] A testcase .fo
Applied, thanks! Glen
batch transition pre-FOP1.0 resolved+fixed bugs to closed+fixed