Hello, On Sat, 16 Jun 2012 19:05:06 +0200 Michael Biebl <biebl@debian.org> wrote: > > Reassigning it back as it really is a bug in NetworkManager. > I've asked for further justification. > Just saying "really" isn't. > If it is a bug in NetworkManager, then please show me where. auto eth0 #NetworkManager#iface eth0 ... is not a valid syntax. So when we have interfaces 'defined' like this, initscripts' hook thinks we've got all 0 interfaces up so it can start. Of course, this needs to be fixed so it won't even try to do so. But the source of the problem is that NetworkManager was abusing a bug in ifupdown's parser. If you really wanted to 'hide' an interface, you should have commented out all the 'auto' and 'allow-' lines, not 'iface', leaving 'auto' intact, which apparently doesn't work. Also calling ifupdown's hooks at random moments isn't a good idea either. > If you suddenly decide to change the behaviour of ifupdown, then > please co-ordinate such a change and get affected packages fixed > beforehand. And let packages know what they need to change. That wasn't suddenly. It's been documented always but didn't work a little bit as expected. Exploiting a bug in a parser is always bad, and there's no excuse for doing that. I can't possibly know every person who's tried to misuse this software, and that's really a problem of those persons. Or shouldn't bugs get fixed at all then if anyone exploits them all the time? I think that reassigning a bug to network-manager in a first place was a clear enough message that something needs to be changed, so reassigning it back multiple times isn't a good way of communication either. -- WBR, Andrew
Attachment:
signature.asc
Description: PGP signature