Strange behaviour on port update

[Andrew Hamilton-Wright]

Dear Python Maintainer,

Something seems to be up with respect to the "flavours" within the
math/py-numpy port.  When running portupgrade, and seeing it attempting to
update py36-numpy, I got the output below.  The line "Registering
installation for py27-numpy-1.13.3_3,1" seems very suspicious to me.

I'm going to try a manual rebuild, specifiying the 36 flavour, but thought
that you would want to know.

Thanks,
Andrew.



Installed packages to be REMOVED:
        py36-numpy-1.13.3_2,1

Number of packages to be removed: 1

The operation will free 17 MiB.
[1/1] Deinstalling py36-numpy-1.13.3_2,1...
[1/1] Deleting files for py36-numpy-1.13.3_2,1: 100%
[Reading data from pkg(8) ... - 386 packages found - done]
--->  Installing the new version via the port
===>  Installing for py27-numpy-1.13.3_3,1
actual-package-depends: dependency on /usr/local/lib/libumfpack.so not
registered (normal if it belongs to base)
actual-package-depends: dependency on /usr/local/lib/libatlas.so not
registered (normal if it belongs to base)
actual-package-depends: dependency on /usr/local/bin/gfortran6 not
registered (normal if it belongs to base)
actual-package-depends: dependency on /usr/local/bin/python2.7 not
registered (normal if it belongs to base)
===>   Registering installation for py27-numpy-1.13.3_3,1
*** Error code 70

Stop.
make[2]: stopped in /usr/ports/math/py-numpy
*** Error code 1

Stop.
make[1]: stopped in /usr/ports/math/py-numpy
*** Error code 1

Stop.
make: stopped in /usr/ports/math/py-numpy
** Command failed [exit code 1]: /usr/bin/script -qa
/tmp/portupgrade20180504-5803-3qkv6p env UPGRADE_TOOL=portupgrade
UPGRADE_PORT=py36-numpy-1.13.3_2,1 UPGRADE_PORT_VER=1.13.3_2,1 make
reinstall
--->  Restoring the old version
Installing py36-numpy-1.13.3_2,1...
Extracting py36-numpy-1.13.3_2,1: 100%
** Fix the installation problem and try again.
** Listing the failed packages (-:ignored / *:skipped / !:failed)
        ! math/py-numpy (py36-numpy-1.13.3_2,1) (install error)