Hi Hans, On Wed, May 13, 2015, at 23:39, Hans van Kranenburg wrote: > found 784852 1.5.0-2 > thanks > > Hi Ondrej S, > > On 05/12/2015 04:51 PM, Debian Bug Tracking System wrote: > > Processing commands for cont...@bugs.debian.org: > > > >> close 784852 1.5.0-2 > > Bug #784852 [bird] bird: Use debconf for router id > > There is no source info for the package 'bird' at version '1.5.0-2' with > > architecture '' > > Unable to make a source version for version '1.5.0-2' > > Marked as fixed in versions 1.5.0-2. > > Bug #784852 [bird] bird: Use debconf for router id > > Marked Bug as done > > Would you mind sharing the rationale behind just closing this bug report > without any comment? It might help others to understand.
It should have been closed with 1.5.0-2 upload, but I used the different bug number by mistake. > I just had a look at 1.5.0-2 > > 1. Thanks for including the smaller example config file. I trust that > Ondrej Zajicek will come up with an even better one, which can replace > it later. > > 2. What is your opnion on the previous thoughts about *never* asking to > replace the bird.conf file? The current handling (also with ucf) shows > the same behaviour as the conffile way. I prefer doing thing in standardized way, that's more predictable and already in place. I can think of one problem that's already solved by using ucf, but would have to be handled by the postinst scripts just top of my head: - one of the standardized ways how to disable some service in Debian is to remove it's conffile, but simply checking whether conffile exists or not in postinst script would simply re-add user removed conffile. To fix that you would have to add additional login in postinst to check whether we have this is a new install or whether we are upgrading the package. My take on this is that it's better to use well tested tool. That said - ucf has an option to always retain user configfile by setting UCF_FORCE_CONFFOLD environment variable (I never used it before), so that's something I am going to use. I will also provide some historical md5sum for bird*.conf (jessie and 1.5.0-1), and that should also provide a way how to limit the dialogues to minimum while providing a way to replace the old conffile with a new one if unmodified. > 3. The package is currently broken. Simply installing the package > results in errors now, because "router id 2001:DB8::1;" is not valid > bird configuration. See attached output. Did you test the package before > uploading? I did, but obviously I missed this. > P.S. are you using git://anonscm.debian.org/users/ondrej/bird.git for > packaging? Could you push your changes? Sure, pushed. O. -- Ondřej Surý <ond...@sury.org> Knot DNS (https://www.knot-dns.cz/) – a high-performance DNS server -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
