On 9.1.13 9:36, Marcel Reutegger wrote:
Hi, I was wondering if the use of headRevisionId in CommitCommandInstructionVisitor is really correct... isn't it more appropriate to use the base revision of the associated commit? from my POV it makes more sense to apply instructions on stored nodes based on the base revision instead of the head revision. I think this also explains the behavior we currently see as described in OAK-507. instead of applying the instructions on the revision passed in MK.commit(), the visitor uses the head revision where the node is already gone.
See also the Javadoc of MicroKernel.commit: "The implementation tries to merge changes if the revision id of the commit is set accordingly".
Michael
WDYT? regards marcel
