Re: [gtk-vnc-devel] [PATCH 0/7] Optimizations for high-latency X connections

[Anthony L. <an...@co...>]

Daniel P. Berrange wrote:
> On Wed, Dec 17, 2008 at 07:46:37PM -0600, Anthony Liguori wrote:
>   
> The former is securely authenticated and encryption, the latter is often
> not - unless using the GTK-VNC + QEMU  VeNCrypt extension, but I can well
> imagine there's plenty of people who don't do that, particularly if they
> are a SSH GSSAPI+Kerberos enabled environment.
>
> Both are valid deployment use cases and we shouldn't presume one vs the
> other, although obviously we can make recommendations.
>   

Also, this is why things like LBX and NX exist.  X is a terrible 
protocol to use over high latency connections.  These patches may make 
it as usable as TightVNC, but that still has to be pretty terrible.

I don't mind the patches that adjust encoding order or implement new 
encodings.  As long as it's in the spec, and we have evidence (via 
traces) that some servers behave better with a different encoding order, 
that's fine to me.

Introducing low-level X dependencies seems like a bad idea to me 
though.  It makes the code less portable because of platform specific 
hacks and it makes things generally less understandable.  If there's 
wide agreement that this is something we should do, I'll concede, but it 
doesn't seem right to me.

Regards,

Anthony Liguori

> Daniel
>