BIND 10 #69: data source hangs with CNAME loops

[BIND 10 Development]

#69: data source hangs with CNAME loops
--------------------------+-------------------------------------------------
 Reporter:  jinmei        |        Owner:  UnAssigned                   
     Type:  defect        |       Status:  assigned                     
 Priority:  minor         |    Milestone:  03. Authoritative-only server
Component:  Unclassified  |   Resolution:                               
 Keywords:                |    Sensitive:  0                            
--------------------------+-------------------------------------------------
Changes (by jinmei):

  * owner:  jinmei => UnAssigned
  * priority:  critical => minor


Comment:

 Replying to [comment:3 each]:
 > On further thought, I think that latter issue should probably be
 addressed in Message.addRRset(), so I'm going to give this ticket to
 Jinmei to think about...

 I think we should rather keep Message::addRRset() dumb and have the
 calling apps handle such a rare, mostly pathological cases.  Assuming we
 (eventually) do tigher validation on generating data sources, this type of
 duplicates should be very rare.  On the other hand, since the message
 method is called per RRset for every response, it would be a huge waste to
 check this rare event in addRRset().

 In fact, BIND9's message module adtops the same policy, i.e., not checking
 duplicate RDATA in the rendering side of code path (although I'm not sure
 if the performance concern was a reason for that).

 I think we should revisit the data source architecture and query logic
 pretty substantially anyway (see
 https://bind10.isc.org/ticket/70#comment:12), however, it makes sense to
 consider this specific topic in that context, too.  So I'll simply keep
 this ticket open as a minor issue with no assignee.

-- 
Ticket URL: <http://bind10.isc.org/ticket/69#comment:4>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development