Re: [O] [babel, bug?] colnames with a list of columns does not work

[Sebastien Vauban]

Hello Thomas and Rainer,

Rainer M Krug wrote:
> Sebastien Vauban writes:
>>
>> #+begin_src R :rownames yes :colnames '(Lg Nb)
>>   data(iris)
>>   head(table(iris$Petal.Length, iris$Species)[, "setosa"], n=2)
>> #+end_src
>>
>> returns:
>>
>>   |     |  x |
>>   |-----+----|
>>   |   1 |  1 |
>>   | 1.1 |  1 |
>>
>> while I was expecting:
>>
>>   |  Lg | Nb |
>>   |-----+----|
>>   |   1 |  1 |
>>   | 1.1 |  1 |
>
> WHy should it? The org-info manual states:
>
> ,----
> | The `:colnames' header argument accepts the values `yes', `no', or
> | `nil' for unassigned.  The default value is `nil'.  Note that the
> | behavior of the `:colnames' header argument may differ across
> | languages. 
> `----
>
> It says nothing about accepting any other values.
> Unless I am missing something?

Yes, you just show that the documentation is not up-to-date, as that
functionality *is* implemented for most languages.

Doing some bit of archeology, I just found out that:

- Eric wrote a patch to support the above (but it hasn't be applied),

- I (!) even wrote a test of that functionality (for a shell block) in
  `testing/lisp/test-ob.el'.

See https://lists.gnu.org/archive/html/emacs-orgmode/2013-04/msg00527.html:

  ┌────
  │ It looks like ob-R implements its own result table reconstruction
  │ instead of using the general support.  This is because R actually
  │ has a notion of column names and row names internally.  The
  │ implementation in ob-R does not correctly handle specified colnames
  │ as your example shows.
  │ 
  │ The attached patch brings ob-R closer to the using the unified
  │ general table reconstructed used in most other languages, and fixes
  │ your problem mentioned above.  I haven't applied it however, as it
  │ may introduce other bugs related to specifying column names from
  │ within R.  For example, I'm not sure that it will now correctly
  │ apply column names from a table built entirely from within R.
  │ 
  │ Additional testing by someone more familiar with R than myself would
  │ be greatly appreciated.
  └────

Should such someone (more familiar with R) be able to confirm that his
patch work without introducing problems, it could be applied so that
R should behave the same as in most languages...

Best regards,
  Seb

-- 
Sebastien Vauban