These test files are basically intended for development use, to test the
various functions of the library and make sure they do what we expect
them to.  However, they can also be useful in illustrating how the
functions work and what their uses are.

Some, like mkstr, can also be particularly useful on their own in certain
cases.  Note that pretty much all of them use cidr_from_str() and
cidr_to_str() to take their input and show their output, of course.

Note that there's a "run.sh" program provided in this dir; this uses the
LD_LIBRARY_PATH environmental variable to allow running the programs
using the shared library, even though it's not in any of the standard
system library paths.  Thus, you'd build the programs, then run them with
an invocation something like "../run.sh ./cidr_mkstr 1.2.3.4/5".


Here's a quick rundown of what they cover:

- regression/ contains a perl script intended for automated regression
  testing.  Currently, it only tests 'mkstr', since the place I'm worried
  about regressions (and finding always more weird edge cases in) is in
  cidr_from_str(); the rest of the functions are sufficiently simple that
  I've not yet found a need for automated testing of them.
  cidr_from_str() is fiendishly complicated, and fixing one thing often
  breaks another, so it's needed here.

- compare/ takes two blocks on the command line, and uses cidr_contains()
  to tell you whether they're the same block, one contains the other, or
  they're overlapping.

- inaddr/ uses the C library inet_pton() and inet_ntop() functions to
  double-check cidr_{to,from}_inaddr() and cidr_{to,from}_in6addr() and
  make sure they act right.  Note that some failures aren't necessarily a
  problem, since in[6]_addr only store addresses, not netmasks, so it's
  not surprising if things don't match sometimes.

- kids/ uses cidr_net_subnets() to recursively show all the ways a given
  netblock can be subnetted.  For safety, this only accepts IPv4 prefixes
  /24 and longer and IPv6 prefixes /120 and up; otherwise it could be
  printf()'ing from now until doomsday.  Trying to show every possible
  subnet of ::/0, for instance, would take JUST a little time (the number
  of output lines would be on the order of 2**129; put THAT through your
  sliderule).

- mkstr/ tests cidr_from_str() and cidr_to_str() by taking in subnets
  from the command line, turning them into CIDR structs, and then
  spitting out string forms of the structs.  What makes it particularly
  useful are the wide array of arguments it accepts to try out various
  combinations of the cidr_to_str() flags and get all the different
  string representations libcidr can create.  It's handy if you can't be
  bothered figuring out what an IPv6 address would look like in fully
  expanded form manually, for instance.  The cidrcalc example program
  shows a lot of the highlights, but this lets you fudge it just about
  any way the library supports.

- netbc/ uses the cidr_addr_network() and cidr_addr_broadcast() functions
  to show the (take a wild guess) network and broadcast addresses of the
  given netblock(s).

- nums/ calls cidr_numaddr() and cidr_numhost() to show the number of
  total addresses and host addresses in the given blocks.

- parent/ keeps calling cidr_net_supernet() to find the parent network of
  the given subnet, and the parent network of that, and the parent
  network of that, and...   all the way up to 0/0 (rather, the v4 and v6
  equivalents of 0/0).