Different results in MiNiFi-CPP

[Marc <ph...@apache.org>]

Hello,

TL;DR Error happens. Error bad. Cookies good.

  I'm getting different results when running Linux vs Java version of
MiNiFi-CPP from head. One major difference is a failure in libUUID when
performing a uuid_compare ( via ProcessorGroups ). I don't always see this
failure -- by way of a segfault. The code referenced by [1] also segfaults
on my machine. I changed one part of MiNiFi-CPP previously to include a
uuid_parse before a comparison, which succeeds as the example in [1] shows.
I discovered this by happenstance. I've found what I think is the offending
code in libUUID, but the problem is not solved in Sierra, and exhibits in
El Capitan, and Yosemite.
    I do not always see this failure on the Mac Sierra; however, my
question is who has seen this error besides me? I've demonstrated this
exact error via [1] on a MBP, and two of my MacAirs, all which different
versions -- they're air gapped from the internets.
    I may be conflating different issues in [1] and MiNiFi-CPP, but the
assembler output is nearly identical so I'm reasonably certain they are the
same issue. Has this demonstrated on others' testing environments based on
your Mac?  Instead of performing a uuid_parse, an alternative may be a
string comparison of two unparsed UUIDs, which does not fail. Before making
such a change I'd prefer to get a cross section of others' experience, to
ensure this simply isn't a function of Marc being Marc.
  Thanks,
  Marc

[1]  https://github.com/lionheart/openradar-mirror/issues/5825