Proposed New Syntax For Calling Code Blocks was: Problem with Babel and sessions

From: "Eric Schulte" <schulte.eric@gmail.com>
To: Christopher Witte <chris@witte.net.au>
Cc: Org Mode <emacs-orgmode@gnu.org>
Subject: Proposed New Syntax For Calling Code Blocks was: Problem with Babel and sessions
Date: Fri, 10 Sep 2010 12:06:47 -0600	[thread overview]
Message-ID: <87vd6deafb.fsf@gmail.com> (raw)
In-Reply-To: AANLkTim32fcEKcLMNV_R6WaUmL=KpojeC8ySaHYyottC@mail.gmail.com

Hi Christopher,

Thanks for the well documented example.

The header arguments on the call line, are actually being used by the
call line itself, not by the R block which is called.  To unpack that,
on evaluation the call line expands to a trivial emacs-lisp code block
which is equivalent to

#+begin_src emacs-lisp :var results=chartPolAngle(pol=tableTestBabel) :session testBabel:file
  results
#+end_src

So the :session header argument is used by the emacs-lisp code block in
the call line, and never propagates to the R code block which is
evaluated.  Notice that header arguments like :results silent, will have
the expected results.

So for now you will have to place the header argument on the actual R
code block.

That said, I think that we should provide a mechanism for passing header
arguments through to called code blocks making the original
functionality you described possible.  I'm not sure how best to do that,
but currently I think we should expand the syntax used to call code
blocks to be more similar to inline code blocks, s.t. header arguments
can be placed inside of an optional square bracket section between the
name of the code block and the arguments.  Using this new proposed
syntax your example below would be written as

#+call: chartPolAngle[:session testBabel :file testBabel.pdf](pol=tableTestBabel)

I'd be interested to hear what the community thinks of this new syntax.

As an unrelated note, we also need to think of a natural way to allow
the header argument portion of a code block to span multiple lines.

Best -- Eric

Christopher Witte <chris@witte.net.au> writes:

> I'm having a bit of trouble with org bable and R. When I try to execute the
> code below (C-c C-c on the call line), as you can see, I get the results
> :nil
>
> ---------------------------------------
> #+tblname: tableTestBabel
> |   0 | 4.40 |
> |  30 | 4.54 |
> |  60 | 7.09 |
> |  90 | 9.40 |
> | 120 | 9.22 |
> | 150 | 6.65 |
> | 180 | 4.22 |
>
> #+srcname: chartPolAngle(pol)
> #+begin_src R
>   angle <- pol[,1]
>   energy <- pol[,2]
>   plot(angle,energy)
>   energy.model <-  nls (energy ~ a1 + a2 * (sin(pi / a3 *(angle-a4)))^2,
> start=list(a1=3.0, a2=7.0, a3=180, a4=0.0))
>   lines(spline(angle, fitted.values(energy.model)), lwd=2)
>   summary(energy.model)
> #+end_src
>
> #+call: chartPolAngle(pol=tableTestBabel) :session testBabel :file testBabel.pdf
>
> #+results: chartPolAngle(pol=tableTestBabel)
> : nil
> ------------------------------------
>
> but If I move the header arguments to the source block, as in the code
> below, everything works fine. I also get asked "ESS [S(R): R] starting data
> directory? ~/doc/org/" which I don't get with the code above.
>
> ---------------------------------------
> #+tblname: tableTestBabel
> |   0 | 4.40 |
> |  30 | 4.54 |
> |  60 | 7.09 |
> |  90 | 9.40 |
> | 120 | 9.22 |
> | 150 | 6.65 |
> | 180 | 4.22 |
>
> #+call: chartPolAngle(pol=tableTestBabel)
>
> #+results: chartPolAngle(pol=tableTestBabel)
> : testBabel.pdf
>
>
> #+srcname: chartPolAngle(pol)
> #+begin_src R :session testBabel :file testBabel.pdf
>   angle <- pol[,1]
>   energy <- pol[,2]
>   plot(angle,energy)
>   energy.model <-  nls (energy ~ a1 + a2 * (sin(pi / a3 *(angle-a4)))^2,
> start=list(a1=3.0, a2=7.0, a3=180, a4=0.0))
>   lines(spline(angle, fitted.values(energy.model)), lwd=2)
>   summary(energy.model)
> #+end_src
> ---------------------------------------
>
> Org doesn't appear to respect the head arguments on the call. I pulled from
> git a couple days ago. Any ideas why?
>
> Cheers
> Chris Witte
> _______________________________________________
> 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