[X2go-dev] [Pkg-x2go-devel] Getting things started with x2goclient

[Gerry Reno]

On 02/02/2011 12:04 PM, John A. Sullivan III wrote:
> On Wed, 2011-02-02 at 10:02 +0100, Erik Auerswald wrote:
> <snip>
>   
>> The problem with NoMachine going closed source is that the people
>> updating, improving and fixing (the free) NX will stop doing so (or
>> have already done that). NX is the enabler for X2Go, but I don't see
>> X2Go having the manpower to tackle the additional task of developing
>> resp. maintaining NX themselves.
>>
>> Erik
>>     
> This concerns us quite a bit.  On the other hand, it plays to one of
> X2Go's strengths.  Heinz and Alex have always maintained that it is not
> a remote protocol solution like NoMachine but a complete Terminal Server
> Project.  Some have criticized it for just being a bunch of wrappers but
> I see that as its saving grace here.  I'd imagine there is no reason why
> something else besides NX could not be slotted into the wrappers in
> place of NX -- perhaps something that handled video and other large
> screen updates better.
>
> Of course, the big question is what.  HP has done some very interesting
> work with adaptive protocols, i.e., they adapt their compression
> algorithms to the needs of the video transfer.  If I understand them
> correctly, they handle the streaming video problem not by spooling the
> file to the physical desktop and playing it locally like Citrix does but
> by adapting the algorithms used for transmitting the video.  I do not
> believe they have open sourced the code.
>
> Almost two years ago, Heinz forwarded me a link to a University project
> that was investigating more video friendly remote video protocols.
>
> So, I don't have an answer but I think we can keep our eyes open for
> something besides NX.  Just an ignorant thought - John
>
>   

John, yes I've also been keeping my eye on some things like this.

There are a number of efforts cropping up that are trying to take
advantage of GPU hardware on the client. 

Wayland by Ubuntu is one that comes to mind. 

The idea is to just send the screen "damage" events down to the client
and let the client GPU hardware handle it.


Regards,
Gerry