On 21/02/2018 01:05, Thomas Huth wrote: > On 20.02.2018 19:37, Paolo Bonzini wrote: >> On 20/02/2018 18:40, Thomas Huth wrote: >>> "-net" is clearly a legacy option. Yet we still use it in almost all >>> examples in the qemu documentation, and many other spots in the network >>> chapter. We should make it less prominent that users are not lured into >>> using it so often anymore. So instead of starting the network chapter with >>> "-net nic" and documenting "-net <backend>" below "-netdev <backend>" >>> everywhere, all the "-net" related documentation is now moved to the end >>> of the chapter. And the examples are changed to use the "--device" and >>> "--netdev" options instead of "-net nic -net <backend>". >> >> Do we want to change them to "-nic" instead? The proof is in the >> pudding, they say, :) and "-nic" is way easier to learn than "-device >> -netdev". > > While -nic is easier to use than -netdev, I don't think that we should > put the focus in our main qemu-doc on -nic instead of -netdev. -nic is a > convenience option, while -netdev is the "architected" way to configure > network devices. We first should document how to do it "right", and > teach the user to proper distinguish between emulated guest hardware and > host network backend (with the old -net command, a lot of people seemed > to have mixed that up IIRC), and then finally explain -nic on top of it.
Heh, that's a philosophy question regarding the organization of the whole manual. Currently the "architected" way is pretty much confined to docs/qdev-device-use.txt. The manual is full of uses of -drive or -hda, and I think it makes sense because honestly that's what users use. I should have explained this in the previous message, sorry. >> And maybe we *should* go the extra mile and deprecate "-net" altogether. >> The only case where the newer syntax is a bit more uncomfortable is for >> "-net nic -net nic -net tap|user", which however does work with "-nic >> hubport -nic hubport -netdev tap|user,id=x -netdev hubport,netdev=x". > > I'd be glad to add such a deprecation patch to this series - I just > thought it might have been too early so far, but if you feel confident > that we can mark it as deprecated, I can spin a v3 with such a patch on > top... I can't deny it's going to be a loooong deprecation. But we have to start somewhere, and -nic is a great start. I think you should send v3 with the minimal changes required to accept these patches, and then leave the rest to a separate submission, but of course you don't have to do it that way. Thanks, Paolo