The only workaround that I could find is to parent the child object to its parent before transforming them (pos=(0,0,0), rot= (0,0,0) scale (1,1,1) for both parent and child). In that particular case, the coordinate shown are the same as they were relative to the parent. But it’s not always convenient, since sometimes I have my objects already transformed.
Anyway, this has been brought into discussion by Sybren, and probably this:
could be a step towards having a different parenting behavior option, I personally have to wait until the option gets implemented and put into a patch to see if that’s what we are asking (my fault, not enough familiarity with the parent inverse matrix),
what do you think about it @Draise ?