RFC: cleanup of linux ports

[Mark Linimon]

I am also still confused by the RUN_DEPENDS logic (some of this logic
is in the current bsd.port.mk).

In certain cases we depend on bin/sh to be the marker, in other cases
etc/redhat-release.  This seems inconsistent.  I wonder if this
would cause problems if trying to install different versions?

This logic also assumes that anything that isn't a partial portname
or '7' should be the default.  This seems wrong.  IMHO after the
test for '7' should be a test for 'yes', which sets whatever is the
current default (currently 8).  Any other value should set IGNORE.
e.g. if someone mistypes USE_LINUX=rh-7 they shouldn't silently get
an install dependency on 8; instead, the installation should fail.
(And either the test should be made sufficiently robust to accept
'y' and 'YES' which occur in 2 ports, or those 2 ports should be
changed to set the value 'yes').

mcl