Re: [Apache Bloodhound] #746: Parent/child relations are inverted in ticket validation (was: Parent/child relations are inverted)

[Apache Bloodhound <de...@bloodhound.apache.org>]

#746: Parent/child relations are inverted in ticket validation
------------------------+-----------------------
  Reporter:  rjollos    |      Owner:  rjollos
      Type:  defect     |     Status:  closed
  Priority:  major      |  Milestone:  Release 8
 Component:  relations  |    Version:
Resolution:  fixed      |   Keywords:
------------------------+-----------------------
Changes (by rjollos):

 * status:  new => closed
 * resolution:   => fixed


Comment:

 (In [1573853])

 0.8dev: Validation that checked for open child tickets had reversed logic.
 Refs #746.

 The validation now checks whether the source ticket //is a parent of// any
 open tickets.

 ----

 After r1573853, there is a failing unit test. This appears to be caused by
 a different defect, described in #775, which was revealed when the unit
 test was fixed in r1573853.

 It appears that the logic of other several unit tests is reversed as well.
 In the method call `RelationSystem.add(source, destination, type)`, my
 interpretation is that the logic should translate to: **source** is a
 **type** of **destination**. So `RelationSystem.add(tkt1, ttk2, 'parent')`
 would add the relation: tkt1 is a **parent** of tkt2. r1573853 corrected
 the logic of the tests `test_cannot_close_ticket_with_open_children`
 according to this interpretation, but several other tests including
 `test_cannot_add_parent_if_this_would_cause_invalid_relations` may also
 need to be modified.

-- 
Ticket URL: <https://issues.apache.org/bloodhound/ticket/746#comment:1>
Apache Bloodhound <https://issues.apache.org/bloodhound/>
The Apache Bloodhound issue tracker