BIND 10 #2272: Improve the perfdhcp command_options_helper code.

[BIND 10 Development]

#2272: Improve the perfdhcp command_options_helper code.
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:  marcin
  marcin                             |                Status:  reviewing
                       Type:         |             Milestone:  Sprint-
  defect                             |  DHCP-20121004
                   Priority:         |            Resolution:
  medium                             |             Sensitive:  0
                  Component:         |           Sub-Project:  DHCP
  Unclassified                       |  Estimated Difficulty:  0
                   Keywords:         |           Total Hours:  0
            Defect Severity:  N/A    |
Feature Depending on Ticket:         |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------
Changes (by stephen):

 * owner:  stephen => marcin


Comment:

 Reviewed commit 3148cf2b018b222e263894b80cd7e92b20c86ceb

 '''configure.ac'''[[BR]]
 Missing semicolon at the end of the "optreset=1" statement in the
 AC_TRY_LINK code for checking optreset. (Although according to my
 config.log, configure added one anyway.)


 '''tests/tools/perfdhcp/tests/command_options_helper.h'''[[BR]]
 The "results" array allocated is too long: the number of elements in the
 array should not be multiplied by "sizeof(char*)".

 The !CommandOptionsHelper class is only used in unit tests using Gtest.
 With this in mind, when allocating memory with "new", rather than using
 "new (std::nothrow)" then an assert() to check that the returned result is
 non-null, why not write a standard "new" within an EXPECT_NO_THROW?


 '''tests/tools/perfdhcp/tests/command_options_helper.cc'''[[BR]]
 > There are at least two ways to reset the state of the getopt() to
 prevent it from trying to access freed buffers. On Linux: the optind must
 be set to 0. This however does not work on BSD and MacOS where there is an
 additional variable declared: optreset. On those systems both optreset and
 optind must be set to 1.
 This seems to imply that both optind and optreset must be set to 1 on BSD
 systems only (i.e. with HAVE_OPTRESET set to 1).  However the code only
 sets optreset according to this preprocessor symbol; the setting of optind
 is determined by the macro !__GLIBC!__.

-- 
Ticket URL: <http://bind10.isc.org/ticket/2272#comment:5>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development