On Fri, Feb 23, 2007 at 01:29:09PM +0000, Richard W.M. Jones wrote: > > Date: Thu, 22 Feb 2007 13:52:28 +0000 > From: "Richard W.M. Jones" <rjones@xxxxxxxxxx> > User-Agent: Thunderbird 1.5.0.9 (X11/20070212) > To: libvir-list@xxxxxxxxxx > Subject: Re: PATCH: Autodetect QEMU version > > Mark McLoughlin wrote: > > Note, this means we don't have support for networks if something's > >wrong with qemu ... think we should just try and locate the binary when > >starting domains and return an error if it fails. > > This is interesting: I've taken the opinion in the remote patch that > locating binaries is something that is done at configure time. The > effect is that libvirt/libvirtd contain hard-coded paths to particular > binaries (/usr/bin/logger in that instance). What do we think about that? For something like the logger I think it is reasonable to hardcode it - it is a core system binary, so not likely to be moved around / replaced. So detecting the path at build time is fine. For 'qemu' the runtime detection is nicer because it gives users more flexibility - if they're playing around with a development build of qemu or some such, they can just update their $PATH & libvirt will find it correctly. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|