https://bugzilla.redhat.com/show_bug.cgi?id=1542666
--- Comment #2 from Laine Stump <laine(a)redhat.com> ---
(1) Okay, I see now that "Requires: perl(IO::Pty)" is properly resolved by
rpmbuild to require perl-IO-Tty. I thought I had tried that, but most likely
only tried with the original "Suggests: perl(IO::Pty)" and incorrectly jumped
to the conclusion that it was due to the module being included in the IO::Tty
package (when in fact it was just because "Suggests:" is a useless directive
:-P). I finally got it working correctly with Requires perl(IO::Tty) and
apparently didn't try backtracking enough.
(2) I haven't been able to figure out a valid use for "Suggests:" Since
it's
not possible for dnf to know how applications I will install in the future
might use Net::OpenSSH, there's no way for the presence of a "Suggests:
perl(IO::Pty)" to ever actually lead to that module being installed - as far as
I can see the only use for Suggests: is to provide a clue to someone whose
application fails, and who then knows enough to go back to the specfile for the
parent package, which seems not much of a use at all.
This is the sum of my usage of Net::OpenSSH (in the libvirt-tck package, which
starts up a throwaway virtual machine to test the libvirt library):
my $ssh = Net::OpenSSH->new($guestip,
user => "root",
password => $tck->root_password());
($stdout, $stderr) = $ssh->capture2("bob");
($stdout, $stderr) = $ssh->capture2("loblaw");
...
Yeah yeah, allowing password authentication is bad security, I get it. I
imagine it's reasonably common though, especially in throwaway scenarios like
this (and I also can't imagine that perl-IO-Tty is very large).
(3) To be honest, I think Suggests: is pointless, and any module that is
referenced within a package should have a hard Requires:. If someone can
explain to me how having Suggests: can lead to the modules being magically
installed once they are called, then I might change my opinion :-)
--
You are receiving this mail because:
You are on the CC list for the bug.