keywords tag files results differ depending on process

[Gérard 91]

HI
I notice four different behavior
A writing IPTC in file (Ctrl I) in KeyWord pane and choosing category (displayed in column) by double click
=> IPTC KW displayed information is a hierarchical one : "grandparent|parent|child"
=> XMP dc "Subject [n]" field display only one subject : "child"
=> XMP lr "hierarchicalsubect [n]" field display all :"grandparent|parent|child"
=> Category : “Child” only is toggled (or not toggled -depends on option box choose*-)
(*)"Import IPTC or XMP Keyword in category” option box, in option menu,

B chosing the same category but using drag and drop file upon "child", in category tree
=> IPTC KW displayed information is a not lr hierarchical i.e. display : " grandparent,parent,child"
=> XMP dc "Subject [n]" field display only one subject ("child")
=> XMP lr "hierarchicalsubect [n]" field display all "grandparent|parent|child"
=> Category: "child" only is toggled but not "parent" and "grandparent"

C chosing the same category but using check box "child" in category tree
=> IPTC KW relative information is a not hierarchical i.e display : "grandparent,parent,child"
=> XMP dc "Subject [n]" fields display:
"Subject [n]" : "grandparent"
"Subject [n+1]": "child"
"Subject [n+2]": "parent"
=> XMP lr "hierarchicalsubect [n]" field display all
"hierarchicalsubect [n]" : "grandparent"
"hierarchicalsubect [n+1] : "grandparent|parent|child"
"hierarchicalsubect [n+2]" : "grandparent|parent"
=> Category "child", "parent" and "grandparent" are all toggled.

D choosing category by right click on a file
=> IPTC KW relative information is a not hierarchical and display only : "child"
=> XMP dc "Subject [n]" field display only "child"
=> XMP lr "hierarchicalsubect [n]" field display all "grandparent|parent|child"
=> Category "child" only is toggled.

This is amazing. Are these differences doing it on purpose ?

I have not yet well-considered what should be ideally, but at least it seems that for IPTC KW field the way of writing should be the same in all case, and different behaviour should be explained if there is good reason to be so.

Thanks again for continuous improvement
Gérard