On Tuesday 18 November 2008 15:21:54 Richard wrote:
On Tue November 18 2008 8:00:37 am Stanislav Visnovsky wrote:
On Saturday 15 November 2008 18:59:33 Richard wrote:
On Sat November 15 2008 8:44:00 am Rajko M. wrote:
On Saturday 15 November 2008 07:43:26 am Dave Plater wrote:
Something like k3b does, if another application like kiocd has control it gives you the option to kill it. Nice, you should put in an enhancement request and post the number so as I can vote.
Not good. Killing another application using software database can be dangerous. Joy of having instantly available Software Management can be spoiled with broken state of the system ;-)
-- Regards, Rajko
Have Yast send a 'signal' to the updater that says 'quit what you are doing and exit'. Then the updater can shut down gracefully. This can be done as part of the Yast message saying it can't start the software management. It is preferable to having the user go to 'top' and killing the updater cold which is what many do, dangerous or not because they don't want to wait for the updater that often seems to take 'days' to complete. When I am at the console doing maintenance on the system, *I* should have the higest priority, after all, in that portion of Yast, *I* am root and root shouldn't have to wait for an automatic program that always seems to be running when one least needs/wants it.
You can already ask packagekitd to shut down via dbus. But there is no guarantee that it will really shut down. Killing it may lead to a system corruption as mentioned elsewhere.
Stano
... AS CURRENTLY IMPLEMENTED, that may be true, but the thrust of this is that the updater (*any* program or service for that matter) should be implemented in a way so as to shutdown gracefully if the ROOT so dictates. It would be dangerous for the root to issue an 'erase everything' command also, but as the root he is wearing the 'god' hat (little 'g') and unlike Windoze, must be responsible for the operation and health of the system, NOT Redmond or other entity.
If the updater and Yast can 'talk' and coordinate more efficiently, the the perils of a sudden, cold shutoff are reduced or eliminated. The root is, or should be, willing to take the risk that the software might fail. Software does that, you know. The root is taking a risk by running *any* software on the machine, having 'friendly' software just makes it less risky.
You can ask for shutdown. If it does not suit your needs, you are free to do kill -9, but the risk is clear. The problem for me is how much you want to invest in making everything interconnected, smart and automatic. This is way to hell IMO. We have the basic support for the behavior you ask for, but it may fail, and it will fail for a reason (e.g. running rpm is a good reason to stay away from kill).
As stated, currently the updater can and often does create inordinate delays and the root should not have to wait on a 'service' type program.
It was the case for Beta4, where packagekitd did not shut down properly. The bug was fixed for Beta5. Do you still see those? Stano -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org