Hi again all, Le jeudi, 2 octobre 2014, 18.31:20 Didier 'OdyX' Raboud a écrit : > I've gone further in my investigations and therefore filed these two > bugs:
I've got answers on these two bugs: https://bugs.freedesktop.org/show_bug.cgi?id=84604 Lennart Poettering saying: > So, systemd currently doesn't support this scheme nicely. We should > however. > (…) > Now, there's one problem with this: the latter line will cause the > .socket unit to fail on kernels where ipv6 is not compiled in. The > question now is what to do about this. https://www.cups.org/str.php?L4491 Mike Sweet saying: > OK, systemd's IPv4 over v6 address is NOT OK. v4-over-v6 is widely > seen as a major security problem, which is why CUPS doesn't support it > transparently. > (…) > The latest changes for STR #4497 no longer use systemd to make the > localhost listeners, so this won't be an issue anymore... The answers combined on both bugs lead me to the conclusion that the socket-activation of CUPS on its INET ports is too premature in absence of the hypothetical new "failure to listen on a port is non-fatal" systemd feature, which is unlikely in Jessie. I will proceed with dropping the "systemd socket activation by-default" feature in Debian's CUPS in the next days, depending on the actual testing migration. I'm not overly happy with this, but I think this is the best way for a rock-solid CUPS in Jessie. Cheers, OdyX -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org