I believe some of the inter-process communication magic depends upon i/o through a file that exists in /tmp on the remote host
If the remote server shares a file system with localhost (does it?), you could contrive for that temp file to live there (~/.emacs.d/ESS.IPC.HACK) instead. I've been stung by similar in the past. What version of ESS and org are you using? (meta-x org-version and meta-x ess-version) - this has been a bit of a moving target. I've most recently had tramp/ess/org work FINE. That is when I am editing a file on a remote host via tramp. The R process starts up on the remote host (as does a shell for "#+SRC sh" blocks). This _should_ work for you.... ~Malcolm > -----Original Message----- > From: Emacs-orgmode [mailto:emacs-orgmode- > bounces+mec=stowers....@gnu.org] On Behalf Of Charles C. Berry > Sent: Thursday, April 21, 2016 5:05 PM > To: Vikas Rawal <vikasli...@agrarianresearch.org> > Cc: org-mode mailing list <emacs-orgmode@gnu.org> > Subject: Re: [O] Using tramp to run R source code blocks on remote server > > On Thu, 21 Apr 2016, Charles C. Berry wrote: > > > On Thu, 21 Apr 2016, Charles C. Berry wrote: > > > >> On Thu, 21 Apr 2016, Vikas Rawal wrote: > >> > >>> I am trying to run R source code blocks on a remote server (defined in my > >>> ~/.ssh/config as cesp). > >>> > >>> I am facing several problems. > >>> > >>> 1. The following sample block, when used with ":results value" does not > >>> give any results. > >>> > >>> #+NAME: level1 > >>> #+begin_src R :results value :exports results :colnames yes :hline yes > >>> :session cesp :dir /cesp:/home/vikas/ > >>> c(1:5) > >>> #+end_src > >>> > >>> I get the following message: > >>> > >>> tramp-flush-directory-property: Wrong type argument: arrayp, nil > >>> > >> > >> I think this is an issue with having different version of ESS on your > >> local > >> and remote systems. > >> > > > > OOPS! > > > > The `different versions' issue affected my setup. But after correcting it, > > I > > still get the error you cited above. > > > > Not sure exactly where the fault lies. > > > > > Maybe in `org-babel-comint-eval-invisibly-and-wait-for-file', but I am not > sure what/how. > > I tried to instrument the function (C-u C-M-x on the function source) and > run it. And it ran without error when I stepped thru it. Then I removed > instrumentation with C-M-x and reran - again no error. > > I tried removing the *.elc and restarting emacs, but that seems to make no > difference - the error is still there till I instrument the code. > > Until this is sorted out, you might try the instrument/uninstrument gambit > as a workaround. > > HTH, > > Chuck >
