[R] R and DDE (Dynamic Data Exchange)

[Philippe Grosjean]

Richard M. Heiberger wrote:
> I am thrilled to learn tcltk2 has DDE capability.
> It is the piece I have been needing to make ESS work directly
> with the RGUI on Windows.  GNU emacs on Windows has a ddeclient,
> but no access to COM.  So if R, or tcltk2 talking in both directions to R,
> has a ddeserver, all should be possible.  I will be reading the documentation
> closely in a few weeks to tie it together and then intend to make it happen.
> 
> Do you, or any other list member, have a sense of the size, complexity, ease,
> magnitude of the task I just defined?  Any advice as I get started on it?
> 
> Rich

Well, to be honest, DDE is an old exchange protocol (the first one 
proposed by M$ in Windows version 1 or 2). It is not that reliable. In 
practice, when the communication is working fine, you have no problems 
with it. But if something fails in either the server or the client, you 
got a very bad behaviour sometimes.

I think there is some interest to have DDE available for R (WinEdt uses 
DDE, I think... Uwe???), together with (D)COM, and socket server. 
Currently, I am improving the socket server build in svSocket (SciViews 
bundle) because it is the communication protocol we decided to push 
forward in Tinn-R, but there are other implementations out there. I 
think that using a socket server is more reliable and it is also a 
cross-platform solution. So, I would personnally prefer that solution.

Best,

Philippe Grosjean