On 2 January 2011 23:36, Mario 'BitKoenig' Holbe <mario.ho...@tu-ilmenau.de> wrote: > Well, setting up DEBIAN_SNORT_OPTIONS from two variables would probably > also be an option. And probably a bit easier :) However, it's your > package... :)
Yes, that's one I'm thinking of doing anyway, but did not want to introduce just yet to minimize the changes and make it possible for this package to go through the freeze. > Well, at least I'm quite sure it was like this here at my system. And > I'm also quite sure I never touched snort.debian.conf, because I know > it's under postinst control (maybe it was even me who once created that > shell-snippet, don't remember anymore). Here is the output of my first > upgrade attempt (from 2.8.5.2-5): (...) It might be that when this bug was around (#608635) the conffile handling did not work properly. It probably did not automatically detect that the file had not been changed by the user, and attributed the '-p' vs '-p -p' difference to a local change when it was really a postinst change. Now, with this bug fixed it should be transparent to users that have snort/disable_promiscuous to true. If you can, you could purge snort, downgrade to 2.8.5.2-2 (in testing) and then upgrade to 2.8.5.2-7 (reconfiguring snort in the process) to verify this. > No need to mention: the strange tmp-file shouldn't have been lingered > around :) Yes, I noticed this too. It is already fixed in the -7 version :) > Well, your solution, for example, a) makes it hard for users to even > notice there was a change that has been abandoned (I just noticed it > because I'm paranoid), and b) even if they *did* notice it, it's not > that easy to migrate their own file to the new system - they have to > read your postinst code to do that (at least this was what I stuck at). > This is btw. the same with x11-common, where they did also re-invent the > wheel - which you are using now :) Well, migrating to the new configuration handling is easy, just run 'dpkg-reconfigure snort'. The configuration file is not changed in upgrades if you modify it but it is changed if you run dpkg-reconfigure (and its checksum stored). The header of the new configuration files say this now and for some users it might work easily since the config script preseeds your values to debconf. This will work at least for those that have not heavily changed the file, and have just changed the values in the options. In any case, I agree with you that ucf should be a better alternative that the code that is right now in the postinst. Regards Javier -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
