BIND 10 #444: MemoryZone class

[BIND 10 Development]

#444: MemoryZone class
------------------------------+---------------------------------------------
      Reporter:  vorner       |        Owner:  jinmei   
          Type:  task         |       Status:  reviewing
      Priority:  major        |    Milestone:           
     Component:  data source  |   Resolution:           
      Keywords:               |    Sensitive:  0        
Estimatedhours:  0.0          |        Hours:  0        
      Billable:  1            |   Totalhours:  0        
      Internal:  0            |  
------------------------------+---------------------------------------------
Changes (by vorner):

  * owner:  vorner => jinmei


Comment:

 Replying to [comment:2 jinmei]:
 > Question: why do you use map for storing RRsets?  Since the number of RR
 types for a single domain name should normally pretty small, I suspect the
 benefit doesn't away its overhead.  I wouldn't oppose to the use of map
 though - this is a short-to-middle term quick implementation anyway, and
 we'll eventually need full optimization probably with a totally different
 data structure, so minor design issues like this don't matter much.
 Adding some comments about the choice would be helpful though.

 I added the comment. I use it not because of the performance (I guess this
 will not be the bottleneck, as you say, there are only few RRsets, it
 should be marginal difference), but because of interface. With vector, I
 would have to traverse and search it myself. This way I just get the
 answer.

 > Minor nits:
 >  - copyrights seem odd:
 >  I guess it's supposed to be either a single line of ISC copyright or a
 multi-line copyright for both CZ NIC and ISC.

 Hmm, right. I did not care that much care about the headers.

 >  - (not due to this branch but for cleanup) we don't need this doxygen
 markup ("\c"):

 ACK

 >  - MemoryZone documentation: now it has been moved to a different file,
 the doc should be updated accordingly:

 Right, I removed that one.

 >  - I see no need for new tests with the change, but maybe we should move
 some unittests to new files?

 I noticed only when implementing the second task (find) that it does not
 have any tests and wrote a test for the minimal functionality in this
 class at the moment, so the commit was in the middle of the find task. I
 imported that one here, as this is its proper place.

-- 
Ticket URL: <http://bind10.isc.org/ticket/444#comment:3>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development