BIND 10 #1604: reorganize RRset class hierarchy to allow further optimization

[BIND 10 Development]

#1604: reorganize RRset class hierarchy to allow further optimization
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:  stephen
  jinmei                             |                Status:  reviewing
                       Type:  task   |             Milestone:
                   Priority:  major  |  Sprint-20120207
                  Component:         |            Resolution:
  libdns++                           |             Sensitive:  0
                   Keywords:         |           Sub-Project:  DNS
            Defect Severity:  N/A    |  Estimated Difficulty:  5
Feature Depending on Ticket:  auth   |           Total Hours:  0
  performance                        |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------

Comment (by jinmei):

 Replying to [comment:9 vorner]:

 > > > First, I'm not sure it is the best thing to do to throw
 !NotImplemented on getRRsig
 > > This is only in the BasicRRset class.  The idea here is that as
 BasicRRset knows absolutely nothing about signatures (these are introduced
 in the RRset subclass), if you're referencing one and using a BasicRRset
 you've made a mistake.
 >
 > Not necessarily. Imagine a simple datasource backend (let's say the
 built-in one). It makes sense for it to return BasicRRset objects, because
 these RRsets will never be signed. So, from the point of the datasource it
 makes sense this way.

 It seems deeply related to whether we include the concept of signature
 in the highest level of abstraction.  If this discussion is only about
 a short term adjustment (as part of gradual refactoring) to support
 the coming performance work, I think we can adopt either way.

 For a longer term, I personally think it's better to include the
 concept of the signature at the top level, and when we do this it's
 better to have its "get signature" method return NULL when the
 implementation doesn't know anything about signatures rather than
 exception.

-- 
Ticket URL: <http://bind10.isc.org/ticket/1604#comment:10>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development