BIND 10 #678: UDPServer and TCPServer classes need clenup and tests.

[BIND 10 Development]

#678: UDPServer and TCPServer classes need clenup and tests.
-------------------------------------+-------------------------------------
                 Reporter:  vorner   |                Owner:  hanfeng
                     Type:  defect   |               Status:  reviewing
                 Priority:           |            Milestone:  A-Team-
  critical                           |  Sprint-20110316
                Component:           |           Resolution:
  Unclassified                       |            Sensitive:  0
                 Keywords:           |  Add Hours to Ticket:  0
Estimated Number of Hours:  0.0      |          Total Hours:  0
                Billable?:  1        |
                Internal?:  0        |
-------------------------------------+-------------------------------------
Changes (by vorner):

 * owner:  vorner => hanfeng


Comment:

 Hello

 First off, I fixed some codestyle issues (wrapped long lines, removed some
 whitespace at the end of lines, etc) and changed the port number to the
 one we use in other tests (which is above 16k, so it should be generally
 free).

 And I have some comments about the change itself.
  - Timeouts in tests are generally bad and these are pretty long ones,
 slowing down the whole test process. Would it be possible to do it without
 them or at last have them shorter?
  - If it fails the test, it will block forever. That isn't good for our
 automated tests. Would it be possible to abort the test if it doesn't
 terminate in some reasonable time, let's say by alarm() call (called in
 the test constructor and removed in the test destructor)?
  - The tests look quite heavy-weight (eg. quite a lot of code). Not that
 I'd have an idea how to simplify/shorten it, but it would be nice, if
 reasonably possible.
  - Your code seems to be based on something slightly out of date. The
 cycle inside the operator (), checking for while (ec) already handles
 errors (not only bad socket ones) in master, based on the bugfix #657.
 Would you mind merging these together and resolving the collisions that
 will happen there?
  - What does „default“ in changelog entry mean? And it is not true the
 interface changed (the interface is the same, the internal handling
 changed), and as there are mostly tests added, is the user possibly
 interested in this? Is it worth a changelog entry?

 Thanks

-- 
Ticket URL: <http://bind10.isc.org/ticket/678#comment:4>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development