qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Spice-devel] [Qemu-devel] spicevmv chardev, guest agents and paravi


From: Anthony Liguori
Subject: Re: [Spice-devel] [Qemu-devel] spicevmv chardev, guest agents and paravirtual mouse
Date: Wed, 12 Jan 2011 13:36:06 -0600
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.15) Gecko/20101027 Lightning/1.0b1 Thunderbird/3.0.10

On 01/12/2011 01:28 PM, Alon Levy wrote:

Regarding wire protocol this would of course be very wasteful. If it's used
between the guest and host except for the translation overhead, which is minimal
I think (except for a copy-paste of a large amount of data - I'm not familiar
enough with XML-RPC but I guess it has some way to pass binary buffers 
unmodified?),
I don't have a particular objection, other then hoping we consider alternatives 
(there
are a few).

Regarding one-off protocols in general, I agree it creates code duplication
and should be avoided. In that respect I prefer mechanisems that produce
implementations from a declarative description. But I would prefer a binary
protocol, possibly a convertion of the existing spice protocol to declarative
form (like Alex did to the core spice protocols, see spice.proto and 
spice1.proto,
http://cgit.freedesktop.org/spice/spice/tree/spice.proto
Of course there are many implementations of this outside of spice, like protocol
buffers). I also don't think a one-off is necessarily a bad thing.

Yeah, it might be worthwhile to look at protocol buffers.

Regards,

Anthony Liguori

Note while on the subject of design, I think that having some sort of
capabilities negotiation so that we can provide compatibility between
different versions is important.

Regards,

Hans





reply via email to

[Prev in Thread] Current Thread [Next in Thread]