@Ken: ordinary bugs are always fixed in source, so they only affect the development, next-to-be release (currently Trusty). This particular bug was already fixed in Saucy (13.10), because ia32-libs was dropped there.
Only critical bugs warrant an SRU to be backported to, say, Precise. And this is not the case with this bug. Actually, this is not a bug at all, since you can easily not pull boinc-client or ia32-libs by installing boinc-manager with --no-install-recommends.
IMHO, this should be closed as "Fix Released" (since it is "fixed" from Saucy onwards), or even "Invalid", since recommended packages are optional.
@LocutusOfBorg: libboinc7 does not depend on boinc-client, in any Ubuntu release.
I think the only real bug is that boincmgr (the executable, not the package) needs to be run in the boinc-client data dir (/var/lib/boinc-client), and that dir does not exist if boinc-client was not installed. Creating a dummy one at install time (via .dirs or postinst) is not so simple: it requires boinc:boinc perms, a user/group that is only created by boinc-client. Also I'm not sure what would happen if later someone install boinc-client: 2 packages that both create (and thus "own") the same dir and user? How would apt deal with it?
So the best solution would be for boincmgr source code not to depend on datadir at all. It could check its existence and, if not found, assume a "clientless, remote-only" usage and act accordingly. This would be outside the scope of *this* but, but it fits debian's http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=730332
@Ken: ordinary bugs are always fixed in source, so they only affect the development, next-to-be release (currently Trusty). This particular bug was already fixed in Saucy (13.10), because ia32-libs was dropped there.
Only critical bugs warrant an SRU to be backported to, say, Precise. And this is not the case with this bug. Actually, this is not a bug at all, since you can easily not pull boinc-client or ia32-libs by installing boinc-manager with --no-install- recommends.
IMHO, this should be closed as "Fix Released" (since it is "fixed" from Saucy onwards), or even "Invalid", since recommended packages are optional.
@LocutusOfBorg: libboinc7 does not depend on boinc-client, in any Ubuntu release.
I think the only real bug is that boincmgr (the executable, not the package) needs to be run in the boinc-client data dir (/var/lib/ boinc-client) , and that dir does not exist if boinc-client was not installed. Creating a dummy one at install time (via .dirs or postinst) is not so simple: it requires boinc:boinc perms, a user/group that is only created by boinc-client. Also I'm not sure what would happen if later someone install boinc-client: 2 packages that both create (and thus "own") the same dir and user? How would apt deal with it?
So the best solution would be for boincmgr source code not to depend on datadir at all. It could check its existence and, if not found, assume a "clientless, remote-only" usage and act accordingly. This would be outside the scope of *this* but, but it fits debian's http:// bugs.debian. org/cgi- bin/bugreport. cgi?bug= 730332