Re: [O] [PATCH] -for review- Tangling with variable transfer of variables

[Charles C. Berry]

On Fri, 20 Jun 2014, Rainer M Krug wrote:

> Attached please find =the reworked patch.
>
> 1) uses local() and closes connection
> 2) does not leave a variable cal;led file behind

Seems to work on a few test cases.

I'd move the strings to defconsts.

Putting a newline before each `%S' will outdent the first line of 
:var value when it is echoed in the session log so it will be easier to 
view.

===

With Eric's input on this:

Would you also change this at the end of org-babel-R-assign-elisp:

                        res})" name file header row-names max))))
-    (format "%s <- %s" name (org-babel-R-quote-tsv-field value))))
+    (format "%s <- %S" name
+           (if (stringp value) value (prin1-to-string value)))))

??

I think this is innocuous for plain strings with no internal quotes, but 
allows almost anything to be passed to R as a string.

So
        :var x=[1 2 3 "4" 5]

will be passed to R as x <- "[1 2 3 \"4\" 5]"

instead of x <- "[1 2 3 "4" 5] which throws an error.

And this just works:

#+NAME: latex-capt
#+BEGIN_SRC latex
  \begin{math}
    y = X\beta,
    \mbox{where } \beta = %beta%
  \end{math}
#+END_SRC

#+BEGIN_SRC R :var ytxt=latex-capt :results raw :wrap latex
beta <- 1.234
sub("%beta%",beta,ytxt)
#+END_SRC


HTH,

Chuck