On Sun, 03 Apr 2011 01:59:02 +0530 Josselin Mouette <joss@debian.org> wrote: > Le jeudi 31 mars 2011 à 09:25 +0200, Vincent Danjean a écrit : > > Martin F. Krafft started to implement a replacement of ifupdown that > > is better designed. But, due to lack of manpower I think, this project > > did not finish. See this archives of netconf-devel@lists.alioth.debian.org > > for more info. > > I wonder what amount of features we are missing for network-manager to > do the job; instead of rewriting a daemon from scratch, we might as well > use one that was designed mostly for the same purpose. It’s > event-driven, it’s extensible, and its features list is already > impressive. Although it has some bugs remaining to fix, this would also > be the case of the new implementation. > > The primary drawback I see is that some people object to having D-Bus > installed on their systems. But should we manage to get NM being the > default, we could keep ifupdown in the archive to manage trivial setups > with less disk/memory usage; just as an optional replacement. I don't consider network-manager suitable for this purpose - wicd is an alternative to network-manager but I'd hesitate before considering wicd as a replacement for ifupdown. If it came down to a choice between no ifupdown and choosing either wicd or network-manager for systems which currently use ifupdown, we'd simply have to reinvent ifupdown or use static configuration. Principle problem with either wicd or network-manager is handling USB networking on embedded devices. ifupdown has problems (tends to need ifdown before ifup) but we can live with that because it has no extraneous dependencies (it's doesn't depend on dbus or anything not already installed in a system based only on Priority:required). That is my main objection: A replacement for ifupdown *must* not depend on anything which is not already in Priority:required, with the possible exception of net-tools or something similar which itself doesn't depend on stuff outside required. Especially: no dbus requirement, no policykit requirements, no implicit expectation that wireless must be supported on all installations, no requirement for all installs to have encryption support of any kind. Wireless is common, yes, but it is NOT ubiquitous, especially in the embedded space. Any replacement must, to me, only have wireless support as an optional add-on. wicd is ruled out because it's Python, it also has similar expectations to Network-Manager that wireless is mandatory. That's fair enough for what wicd tries to do, it just means that it is not a replacement for ifupdown. Network-Manager is ruled out because it expects every install to need to cope with wireless and therefore brings in (currently) gnutls, gcrypt, tasn, polkit & wpasupplicant type stuff AS WELL as dbus. -- Neil Williams ============= http://www.linux.codehelp.co.uk/
Attachment:
pgpywNPEv4SVm.pgp
Description: PGP signature