BIND 10 #1299: compare SOA serial in xfrin

[BIND 10 Development]

#1299: compare SOA serial in xfrin
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:  jinmei
  jinmei                             |                Status:  reviewing
                       Type:  task   |             Milestone:
                   Priority:         |  Sprint-20111220
  blocker                            |            Resolution:
                  Component:  xfrin  |             Sensitive:  0
                   Keywords:         |           Sub-Project:  DNS
            Defect Severity:  N/A    |  Estimated Difficulty:  3
Feature Depending on Ticket:  AXFR-  |           Total Hours:  0
  in                                 |
        Add Hours to Ticket:         |
                  Internal?:  0      |
-------------------------------------+-------------------------------------

Comment (by jinmei):

 Replying to [comment:14 jelte]:

 Thanks for the review.

 > I fixed a couple of typos, see commit TODO

 Thanks, they look good.

 > xfrin.py.in:
 >
 > in `__parse_soa_response`:
 > {{{
 >             raise XfrinProtocolError('Invalid number of questions to ' +
 >                                      'SOA query (' + str(n_question) +
 ')')
 > }}}
 > While correct, I fear that this message might prove unuseful; we are
 talking about a response where, so I suggest making it something like
 "Invalid number of records in question section of response to SOA query",
 or something.
 >
 > Similar issue for the next exception 'Questions mismatch to ')

 I see the point.  I've updated them in the latest commit.  Hopefully
 they are better.

 > (the rest seems fine in this regard, although the 'at all' in 'SOA query
 resulted in no SOA at all' might only make sense in the context of the
 other possible errors :) I'd suggest 'No SOA record found in response to
 SOA query')

 Ack.  Applied in the same commit.

 > <log-level-bikeshed> should XFRIN_ZONE_SERIAL_AHEAD be a WARN instead of
 INFO? (no strong opinion, just wondering, it is something that should
 probably be looked in to) </log-level-bikeshed>

 In general, I personally believe WARN/ERROR is too verbose for events
 that are triggered outside of the system.  So my general opinion
 applied here.

 It's also consistent with BIND 9's policy:
 {{{#!c
                         dns_zone_log(zone, ISC_LOG_INFO, "serial number
 (%u) "
                                      "received from master %s < ours
 (%u)",
                                      soa.serial, master, oldserial);
 }}}

 I've been aware that the policy on the verbosity level is not very
 consistent among developers.  Maybe a topic for the biweekly call?

-- 
Ticket URL: <http://bind10.isc.org/ticket/1299#comment:15>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development