BIND 10 #2340: build failures with newer versions of clang++

[BIND 10 Development]

#2340: build failures with newer versions of clang++
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:  jinmei
  jinmei                             |                Status:  reviewing
                       Type:         |             Milestone:
  defect                             |  Sprint-20121023
                   Priority:         |            Resolution:
  medium                             |             Sensitive:  0
                  Component:  build  |           Sub-Project:  DNS
  system                             |  Estimated Difficulty:  3
                   Keywords:         |           Total Hours:  0
            Defect Severity:  N/A    |
Feature Depending on Ticket:         |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------

Comment (by jinmei):

 Replying to [comment:5 jelte]:
 > Regarding the change in memory_datasrc_link.cc, I see why it would work,
 but I do wonder if there is also not some static check that flags using
 methods or constructors that don't put the result somewhere :) (of course,
 if so, it's easy to fix as well)

 I think such warnings normally doesn't happen because there's always
 possibility that a call to the constructor has a side effect (but of
 course there can be a tool that even checks the internal of the
 constructor).  In this particular case, I ran the branch on buildbots
 (including the one running cppcheck, I believe) and didn't get a
 failure report; cppcheck also passed locally.

 So, at least for now I think it's okay to keep this code.  As you
 said, if we encounter a new alarm due to that, we can then consider
 how to work around it.

 > The problem in the scoped_lock one really does seem like a checker bug
 btw
 >
 > Changes look OK, but I must admit I have not tried building it with
 clang myself yet.
 >
 > In effect, the changes to configure.ac are small, but shall we run it
 through buildbot first just in case?

 That was indeed a good idea, and there was indeed a bug identified by
 a `SunStudio` build.  It's fixed at b10b39f.  I also merge a newer
 version of master to this branch.  With these it passed on all
 buildbots.  So I believe it will pass on your environment, too.

 Due to the merge with master, getting the diff is difficult, but the
 only change I made on the branch is b10b39f.  So could you just check
 that particular commit?

 Regarding [clang-build], yes I know that, it's due to an earlier local
 attempt.  I didn't touch them as I thought the difference is minor,
 but if you want I'll override it by git rebase.

-- 
Ticket URL: <http://bind10.isc.org/ticket/2340#comment:7>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development