Jóhann B. Guðmundsson wrote, On 11/17/2009 03:07 AM:
First of all this is not the right place to discuss this..
Secondly if an update of an application requires that application to be restarted after being updated, the updating application should just notify the user that there is an update to an application he's currently using that requires him to quit that application for the updater to proceed updating that application. The updater should simply ask him to finish his work then quit the application he's using and rerun updating.
Just my 2 cents..
JBG
By the time the rpm is being processed, it is too late to have a truly interactive wait: right?
how about this ugly hack description/pseudocode : in the prescript if {running instances of Firefox found} then hook each Firefox instance and \ point them at file:///usr/share/FFwarning/FFwarning.html #give an active user half a chance to drop FF. sleep 15 fi
file:///usr/share/FFwarning.html will contain a text warning message and a countdown. the countdown might be implemented as a javascript or as a timed forced load of the next page. The text warning gives the user notice that they have 15...14... to shutdown Firefox before rpm replaces the running binary and risks all the mentioned data corruption/loss.