Hi, Steve Langasek: > I understand that software (especially desktop software) not coping with > online updates is an existing problem.
Not only that. What _do_ you do when version 2 of $SUPPORTING_PACKAGE implements a broken/superseded API or protocol? You limp along; you either cannot start any new $DEPENDING_PROGRAMs or, when the supporter has been restarted, all the dependants instantly break. Unless you restart them too. We do some restarting for libc updates, but in a haphazard way (we don't shut them down _while_ updating, only for some daemons) and for no other package. > The question is whether we - collectively - think moving to offline > updates is an acceptable way to address this problem. The problem is that an offline updater is an attractive nuisance. It solves a specific problem which is now rare, precisely because we don't solve it and therefore spend time and effort to prevent the "offline update" non-solution from spreading. > raise that question is now, /before/ > it becomes an entrenched assumption and leaves our users with no choice but > to use it because their desktops become unusably broken if you apply package > updates while they're running. > The problem is that some already do. Tried updating firefox ^W iceweasel lately? The irony here is that firefox already has a perfectly capable method of realising that there's a newer version available, and of re-exec'ing itself without any data loss. -- -- Matthias Urlichs -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140912081856.gc19...@smurf.noris.de
