Re: org-babel with R, scrolling an inferior ESS process

From: Dan Davison <davison@stats.ox.ac.uk>
To: Erik Iverson <eriki@ccbr.umn.edu>
Cc: emacs-orgmode@gnu.org
Subject: Re: org-babel with R, scrolling an inferior ESS process
Date: Tue, 16 Feb 2010 21:24:14 -0500	[thread overview]
Message-ID: <87sk90vbu9.fsf@stats.ox.ac.uk> (raw)
In-Reply-To: <4B7AC106.5050109@ccbr.umn.edu> (Erik Iverson's message of "Tue, 16 Feb 2010 10:00:06 -0600")

Hi Erik,

[...]
> * Identity function
> #+begin_src R :session :results output silent
>   test <- function(x) {
>     x
>   }
>   test(3)
> #+end_src
>
> If I use C-c ' to open an R code buffer and C-c C-c, I see the same
> behavior as in ESS. However, when I do C-c C-c on the code block *in
> the test.org buffer*, the code is sent to the *R* buffer, but it does
> not scroll.  In the *R* buffer, once I do scroll to the bottom, I see
> the following:
>
>>
> test <- function(x) {
>   x
> }
> test(3)
> 'org_babel_R_eoe'
>
>> test <- function(x) {
> +   x
> + }
>> test(3)
> [1] 3
>> 'org_babel_R_eoe'
> [1] "org_babel_R_eoe"
>
> Can anyone
> 1) replicate that you don't see the scrolling, even with the
> comint-scroll-to-bottom-on-output variable set to 't'?

Yes, this has been on my todo list for a while! Definitely time to fix it.

> 2) suggest a way to get the *R* buffer to scroll to the bottom when I
> execute a source block from an org-mode file to a running *R* session?

It seems to be fixable by changing a couple of save-window-excursions
into save-excursions (patch below). I haven't thought of anything that
this will break, but in any case I suspect that Eric S. will apply the
correct fix to the main git repo soon.

As for the 'org_babel_R_eoe' stuff, that is org-babel internals that
ideally would not be exposed to the user. I suspect that it is possible
to remove it from the comint buffer -- Eric S. is definitely best placed
to comment on the best way of doing that.

I'm sure this is obvious to you, but while we're on the topic: the
reason that the org-babel R comint evaluation behaviour differs from ESS
is because org-babel required a common mechanism that would work for all
the supported languages; we were aware all along that ESS had already
done an excellent job of doing that for R (plus some other languages).

Please do let us know of any other improvements we can make to using
org-babel with ESS. Good integration with ESS has been one of our aims
from the beginning.

Dan


--8<---------------cut here---------------start------------->8---
commit d84c426027653a41ad02992d06d7c151d26d1f4b
Author: Dan Davison <davison@stats.ox.ac.uk>
Date:   Tue Feb 16 21:08:09 2010 -0500

    babel: allow point to move to end of comint buffer

diff --git a/contrib/babel/lisp/langs/org-babel-R.el b/contrib/babel/lisp/langs/org-babel-R.el
index 1fe2826..c3f9e19 100644
--- a/contrib/babel/lisp/langs/org-babel-R.el
+++ b/contrib/babel/lisp/langs/org-babel-R.el
@@ -39,7 +39,7 @@
   "Execute a block of R code with org-babel.  This function is
 called by `org-babel-execute-src-block'."
   (message "executing R source code block...")
-  (save-window-excursion
+  (save-excursion
     (let* ((processed-params (org-babel-process-params params))
            (result-type (fourth processed-params))
            (session (org-babel-R-initiate-session (first processed-params)))
diff --git a/contrib/babel/lisp/org-babel-comint.el b/contrib/babel/lisp/org-babel-comint.el
index d65f054..e693a40 100644
--- a/contrib/babel/lisp/org-babel-comint.el
+++ b/contrib/babel/lisp/org-babel-comint.el
@@ -45,7 +45,7 @@
 body inside the protection of `save-window-excursion' and
 `save-match-data'."
   (declare (indent 1))
-  `(save-window-excursion
+  `(save-excursion
      (save-match-data
        (unless (org-babel-comint-buffer-livep ,buffer)
          (error (format "buffer %s doesn't exist or has no process" ,buffer)))
--8<---------------cut here---------------end--------------->8---


>
> I am surprised by this behavior since I just assumed comint-mode
> should do the right thing, so I'm not sure where the issue is! Even if
> this can't be changed, I think org-babel is going to be a great use to
> me, thank you!
>
> Best Regards,
> Erik Iverson
>
>
> _______________________________________________
> Emacs-orgmode mailing list
> Please use `Reply All' to send replies to the list.
> Emacs-orgmode@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-orgmode
