At 2:28 PM +0100 2/23/07, Piotr Baranowski wrote: >Alan Cox wrote(a): >> On Fri, Feb 23, 2007 at 10:17:23AM +0100, Ralf Corsepius wrote: >>> I disagree - All programs should immediately terminate upon "Ctrl-C". >>> Switching to a different mirror should be done by any arbitrary key. >> >> Someone please fix emacs to terminate immediately on ctrl-C then ;) >> >> Very large numbers of programs override ^C to be an internal interrupt, >> including things like ftp. Others specifically ignore it (try using ssh >> without that) > >I think we can accept such weirdness of some applications. > >Most people will understand WHY ^c is overriden in ssh for example. > >For me after 10 years of linux experience it was great mistery why a hell >does yum override ^c. ... Yum does not override Ctl-Ç. RPM does. Look at the RPM source, say rpmsq.c. RPM does not block the signals, but rather saves them and if a signal has been caught, terminates with extreme prejudice at the end of whatever it was doing. Thus, signals are not processed immediately and always terminate RPM's operation. Not to start any wars here, but I wrote a yum plugin for FC5 (and just updated it for FC6, laziness on my part) that provides, among other things, sane Ctl-C handling (during downloading) in a way I believe safe for RPM. <http://georgeanelson.com/stablemirror.htm> -- ____________________________________________________________________ TonyN.:' <mailto:tonynelson@xxxxxxxxxxxxxxxxx> ' <http://www.georgeanelson.com/> -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
