Re: [TCLUG:12935] Remote X (client/server rant)

[Chris McKinley <lamfada@lugh.net>]

On Thu, 27 Jan 2000, Dave Sherohman wrote:

> But, when you've got a UI, most activities are initiated from the user's side.

right, and the client is actually running on the remote system.  The
display requests are generated by the remote client, and are served by the
X server on the local system.

> 
> > TCP/IP is inherently a peer-to-peer network, and the
> > client/server distinction refers only to who does the connecting and who
> > does the waiting for a connection.
> 
> Aside from confusion with X having the user sit at the server, which then
> talks to the client instead of the other way around, this distinction breaks
> down for a number of applications (most of which have difficulty dealing with
> IP masqing) where the client connects to the server which then may open a
> connection back to the client.  By your definitition, both pieces of software
> are then both clients and servers to each other, even if one is merely
> reacting to requests made by the other.  (It could definitely be argued that
> they have different roles for different transactions, but the overall/general
> identities are not, IMO, so fickle.)
> 

yes, it is possible for two programs to be both clients and servers, to
other programs and to each other, however the XServer does not make
requests.  The X Server fulfills display requests and passes UI event data
to the client.  I think one of the reasons this is confusing is that
generally we view the display as an extension of ourselves, i.e., a tool,
thus we see the display as making requests to the application, however it
is more appropriate to see the display as, i.e., the speaker/microphone on
a telephone.  These things to not actually speak to the person at the
other end of the line, we do.  The speaker/mic serve to convert the sound
waves to voltage changes, and vice versa, and we do not care how they do
it.

Hmm, I could keep going with the metaphor, but I think I will stop there.

-Chris