On Thu, Dec 09, 2010 at 11:52:03PM +0100, Axel Beckert wrote: > Yavor Doganov wrote: > > Any idea what's the right approach to fix this problem? > > Two ideas come to my mind: > > 1. Ignoring the exit code of the init.d script in the prerem.
Something like: ... pgrep -x gdomap || exit 0 #DEBHELPER# ... ? This would solve the issue, I think, and I don't anticipate any negative effects. However, as it looks like a "dirty" solution, I'm somewhat suspicious and not very keen to implement it immediately... Any advice/comments appreciated. > 2. AFAIK the maintainer scripts know if the package is just removed or > upgraded, so maybe calling it in postinst instead of prerm only if > the package is upgraded. No, the daemon must be stopped in prerm, particularly when doing the lenny->squeeze upgrade, because there's a soname change in the library. > BTW: Do you know if that problem is present in Lenny, too? Yes, definitely. (Dist-)upgrading from lenny with the gdomap daemon not running for some reason would reveal the same issue you encountered. > Well, maybe you can put some code handling this issue specificly > before the debhelper snippets and care that they are not run in case > this special case occurs. See above, that pgrep check is the best I could think of (the procps package is guaranteed to be installed on hurd-* and kfreebsd-* as well). But I'm not sure, looks more like sweeping the problem under the carpet to me. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
