https://bz.apache.org/SpamAssassin/show_bug.cgi?id=8031
Bill Cole <billc...@apache.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |billc...@apache.org
--- Comment #13 from Bill Cole <billc...@apache.org> ---
(In reply to Giovanni Bechis from comment #12)
[...]
> As an unprivileged user, regression test works, as root fails.
This is true for multiple tests on various platforms. E.g. on macOS if SA has
been used previously a bunch of spamd/spamc tests fail because there's some
sort of escape from the test environment that grabs the system-wide
welcomelist. Historically I've run into similar problems on FreeBSD. I believe
there's still an open bug that I opened before I was a contributor noting that
'make test' with an '/opt/local' prefix polluted the base OS (/etc).
I don't think we should encourage anyone to build/test SA as root or support
doing so. In fact, I think we should actively discourage it, not because of
anything special in SA but because it's a spectacularly insecure behavior if
you are working with untrusted code that you have not audited.
Yes, I know that's an extreme position which virtually no one actually adheres
to all of the time. Obviously it is a mistake I've made. I strive imperfectly
toward virtue in a world of sin.
I vote for WONTFIX.
--
You are receiving this mail because:
You are the assignee for the bug.
