[xougen] Re: Dealing with images

[Gian Filippo Pinzari]

Hi Kaleb,

On Sunday 07 December 2003 18:15, Kaleb S. KEITHLEY wrote:
> To be a Standard though, you need to write a specification and have it
> go through X.org's standardization process. A proof of concept, in the
> form of an implementation, is useful too.

We have the proof of concept. Should we write such a "X image 
compression and streaming" specification and submit it for approval to 
the X consortium? Such an extension would be useless without clients 
leveraging it. I'm speaking about GNOME, KDE, Mozilla, OpenOffice, 
Evolution and I could name many more. We need the support of the 
X developers otherwise it's better to use our time and resources to 
write a layer that don't need X apps to be aware of it. We already have 
hard time making a living out of writing OSS software, so I don't think 
we'll follow the Rule and spend development time and resources in 
something that, maybe, nobody will ever use. 

Does OSS development always follow the Rules? Fortunately not. 
Look at the XDamage extension. There were some talks and a 
sample implementation from Havoc. It solved a very real problem, 
so everybody agreed on it. Keith Packard wrote the specification, 
a better version of the code and it immediately went into the 
freedesktop X server. As often happens in the OSS world, somebody 
comes with a solution, then the OSS world elaborates it, improves 
it, plugs the holes and makes it a standard. We would like to follow 
the same OSS rule. Unfortunately we are still at the stage that most 
X people seem to completely ignore (or consider not worth of 
mention) the work we have done and the problems we are trying 
to solve. Obviously I think they are wrong, otherwise I would have 
already stopped complaining in mailing lists ;-). I'm just starting 
to believe that also the OSS world suffers of the 'not invented 
here' syndrome.

I carefully read the "Open Source Desktop Technology Road Map" 
of Jim Gettys:

http://freedesktop.org/~jg/roadmap.html

It doesn't make any mention of NX. This is really sad. I'm sure Jim 
knows about our software, so I must argue that he thinks that X 
doesn't need specific X protocol compression, X image encoding 
and streaming, a X agent system resolving round-trips at application 
server side, embedding of different remote desktop protocols in X, 
a proxy system implementing bandwidth control, encryption, 
transport over a RTP network, session initiation through SIP and 
other things we instead need in NX. Should we write a specification 
for each functionality and wait the X developers to embrace them? 
It would be fantastic, but I don't think we have the money to live
long enough to see this happen.

I read in the previous document that Jim Gettys thinks that ssh -X -C 
or VNC are good enough for most remote computing needs. Well, I 
think that the proof of the pudding is in eating. I would really like to 
know if Jim Gettys has ever tried our software and if he had a chance 
to run it, side by side, with ssh -X -C or VNC. 

Many X people seem to forget that there is a company whose name 
is Citrix and another company whose name is Microsoft that have 
nearly the 100% of the remote computing market. Probably these X 
people should rather argue that ssh -X -C is no-good. Some weeks 
ago we received an e-mail from a company that qualified itself as 
a Citrix partner since 10 years. They told us that they had tried NX 
and were stunned by the performances. It was the first time, since 
10 years, that they had something that could beat Citrix. They are 
now in the process of becoming distributors. And, yes, they had 
tried ssh -X -C and VNC. 

I read again the thread about NX in the old XFree86 forum. The point
of Keith Packard and Jim Gettys was that the work that is taking place 
at freedesktop.org should make possible to run remote X applications 
with the same efficiency and with the same features provided by NX
without any of the NX solutions to the problem. I studied the papers 
but I don't see any solution to the "X image encoding and streaming" 
problem, so I think this is a good place to start working together. 
And I hope, Kaleb, that your suggestion is not to do all the work by 
ourselves ;-). 

Kind regards,

/Gian Filippo Pinzari.