tisdag den 18 maj 2010 klockan 12:35 skrev Alberto Gonzalez Iniesta detta: > On Tue, May 18, 2010 at 04:31:11AM +0100, Ian Beckwith wrote: > > Hi, > > > > [MEA wrote] > > > A working xinetd configuration is included. Openbsd-inetd remains > > > untested! > > > > So, if I'm right, the only thing to do is to document that admins who > > want ipv6 support should manually add a tcp6 entry to /etc/inetd.conf.
In NEWS.Debian, or as a simple hook in the postinst script? Like the insert I made in 'postinst' to announce the xinetd example. Just another conditional to catch the proper trigger! > > Oh, and make sure that maintainer scripts don't cause an error with > > this configuration during upgrades. > > Yep. Sad. > I have checked the situation an a system with a clean, new and untouched openbsd-inetd. On first install, ftpd will insert a single tcp-listener, i.e., an IPv4 listening service. On subsequent upgrades (to the identical version), be it with entries flush left, by blank indentation, escaped with hashes, '#<off>#', or '#<off># ', everything goes smoothly with one exception, independently of 'tcp4', 'tcp6' or both at once. The listener or listeners are activated as expected, or are left intact for non-standard indents or escaping. Observe that both or none of 'tcp4' and 'tcp6' are activated simultaneously. By the way, not only is 'tcp46' obsolete, it generates a failure with openbsd-inetd_0.20080125-4. Said exception is that, in case there are two active entries, 'tcp4' and 'tcp6', then update-inetd will launch whiptail with a question as the correctness in deactivating both entries. It does not matter what the answer is, because both stanzas get reactivated by the postinst script in its final step. Could this whiptail dialog (or its counterpart) be of concern in automated installs? > > One more thing, Mats, it occurs to me that your changes (to linux-ftpd > > and netkit-ftp) are significant enough we should clarify the license. > > Good point, Ian. I didn't ask him..... /me runs from debian-legal > > > The current code base is licensed under a standardish bsd-style > > license, copyright Carnegie Mellon University, although of course it > > has been hacked on by a gazillion others over the years. Are you happy > > for your changes to follow the same license? I think you would be well > > within your rights to ask for something like: > > > > "Portions Copyright 2010 Mats Erik Andersson, licensed as above" > > > > in debian/copyright, and/or the source files you changed. > > That looks great. > I took the liberty to read the copyrights for netkit-ftp, netkit-tftp, and linux-ftpd. They are all to my taste, so my contribution can very well be counted as either variant of a BSD formulated license, be it Carnegie Mellon Univ., or Univ. of California. Probably my name belongs in 'debian/copyright' or in the patches themselves, since I undertook the labour to keep the upstream source intact, and only later impose my code additions. So I gather that it is the patches themselves that stem from my thoughts. The notion of 'derived work' escapes my understanding at the moment. Mats Erik Andersson, fil. dr 2459 41E9 C420 3F6D F68B 2E88 F768 4541 F25B 5D41 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
