Re: Babel: communicating irregular data to R source-code block

[tsd@tsdye.com (Thomas S. Dye)]

Eric Schulte <eric.schulte@gmx.com> writes:

> [...]
>>
>> I'm beginning to see why you have strong feelings about python.
>
> Semantically meaningful whitespace is a bad idea for a programming
> langauge.
>

Yes, this makes sense to me.  I suppose I should wean myself from python
now that I use babel as a "glue language."

>> In the code above, the blank line before #+end_src is necessary and
>> must not contain any spaces, and :var n can be set to anything, since
>> it is declared for initialization only.
>>
>> The code in the JSS article doesn't run for me with a recent Org-mode
>> unless I add a blank line before #+end_src, or remove the :return header
>> argument.  If I remove the :return header argument, then the need for
>> the blank line goes away.  The following code block seems to work:
>>
>> #+name: pascals-triangle
>> #+begin_src python :var n=2 :exports none
>> def pascals_triangle(n):
>>     if n == 0:
>>         return [[1]]
>>     prev_triangle = pascals_triangle(n-1)
>>     prev_row = prev_triangle[n-1]
>>     this_row = map(sum, zip([0] + prev_row, prev_row + [0]))
>>     return prev_triangle + [this_row]
>> return pascals_triangle(n)
>> #+end_src
>>
>> #+RESULTS: pascals-triangle
>>
>> | 1 |   |   |
>> | 1 | 1 |   |
>> | 1 | 2 | 1 |
>>
>> I'm guessing that the need for a blank line when using :results has
>> arisen since the JSS article was published, because the article was
>> generated from source code and didn't show any errors.  
>>
>
> I believe that we used to pad code blocks with newlines when they were
> extracted from the buffer, which had the effect of automatically adding
> this extra line.  This behavior however caused problems in some cases
> where the extra line was not desired.
>
>>
>> If I have this right (a big if), then might it be possible to
>> re-establish the old behavior so the JSS code works?  
>>
>
> I've just pushed up a patch in which the addition of the return value in
> python is careful to add this newline itself.  This should restore the
> functionality of the python code from the paper (specifically the
> following now works [1]).  This is applied to the maint branch so
> hopefully it will sync with Emacs before the release of Emacs24.
>

Thanks Eric.  The source block in the paper returns the correct result
with the code in the maint branch.

All the best,
Tom

> Best,
>
>
> Footnotes: 
> [1]  
>
> #+name: pascals-triangle
> #+begin_src python :var n=2 :exports none :return pascals_triangle(n)
> def pascals_triangle(n):
>     if n == 0:
>         return [[1]]
>     prev_triangle = pascals_triangle(n-1)
>     prev_row = prev_triangle[n-1]
>     this_row = map(sum, zip([0] + prev_row, prev_row + [0]))
>     return prev_triangle + [this_row]
> #+end_src
>
> #+RESULTS: pascals-triangle
>
> | 1 |   |   |
> | 1 | 1 |   |
> | 1 | 2 | 1 |

-- 
Thomas S. Dye
http://www.tsdye.com