Re: [BUG][babel] Tangling into directories does not add directories to org file

[Rainer M Krug <Rainer@krugs.de>]

[-- Attachment #1: Type: text/plain, Size: 7007 bytes --]

Grant Rettke <gcr@wisdomandwonder.com> writes:

> Sounds like a valuable use of R.

It definitely is.


>
> Do you have a writeup on the approach or is it a standard feature of
> org and ess?

No - but please bug me - I can post in the intermediate future an
example demonstrating the points.

But the main points are:

* Setup
** I added the following to my emacs.org
*** Add local variable in tangled files if they match [.]R
The variable =org-babel-tangled-file t= enables ess to trace back the
source from the .R file to the .org file, using the comments which
have to include the org links.

See the thread "Automatically adding local variables to tangled file" for details

#+begin_src emacs-lisp
(defvar org-babel-tangled-file nil
  "If non-nill, current file was tangled with org-babel-tangle")

(put 'org-babel-tangled-file 'safe-local-variable 'booleanp)

(defun org-babel-mark-file-as-tangled ()
  (when (buffer-file-name)
    (when  (string-match "[.]R" (buffer-file-name))
      (add-file-local-variable 'org-babel-tangled-file t)
      (add-file-local-variable 'buffer-read-only t)
      (add-file-local-variable 'eval: '(auto-revert-mode))
      (basic-save-buffer))))

(add-hook 'org-babel-post-tangle-hook 'org-babel-mark-file-as-tangled)
#+end_src

** Tangling
- The files have to be tangled with the header argument =:comments link=
(or both) for this to work.
- =:padline no= is necessary (at least it was) as the empty lines in
  the tangled file caused ess to identify the wrong line in the source
  code

** This should now work.

* Use code as "package"
Effectively, I do the following:
** code
tangle all .R files to ./R *without variables transfer*. These should
contain definitions of functions and not executable code. 

** data
data files (i.e. the variables transferred from org) are transfered as
tangled files in in the ./data directory.

*** Species names and espre codes of species used by Vallet
:PROPERTIES:
:header-args+: :tangle ./data/SPECIESVALLET.R
:header-args+: :rownames yes
:header-args+: :colnames yes
:END:
#+NAME: SPECIESVALLET
|         | fullName        | shortName | espre | IFNName         | color |
|---------+-----------------+-----------+---------+-----------------+-------|
| fagus   | Fagus sylvatica | fagus     |       4 | fagus_sylvatica | red   |
| quercus | Quercus robur   | quercus   |       3 | quercus_robur   | green |
#+begin_src R :var SPECIESVALLET=SPECIESVALLET

#+end_src

which evaluates to ./data/SPECIESVALLET.R

,----
| ## Species names and espre codes of species used by Vallet
| ## :PROPERTIES:
| ## :header-args+: :tangle ./data/SPECIESVALLET.R
| ## :header-args+: :rownames yes
| ## :header-args+: :colnames yes
| ## :END:
| ## #+NAME: SPECIESVALLET
| ## |         | fullName        | shortName | espre | IFNName         | color |
| ## |---------+-----------------+-----------+---------+-----------------+-------|
| ## | fagus   | Fagus sylvatica | fagus     |       4 | fagus_sylvatica | red   |
| ## | quercus | Quercus robur   | quercus   |       3 | quercus_robur   | green |
| 
| ## [[file:prodMixStands.org::*Species%20names%20and%20espre%20codes%20of%20species%20used%20by%20Vallet][Species\ names\ and\ espre\ codes\ of\ species\ used\ by\ Vallet:1]]
| SPECIESVALLET <- local({
|      con <- textConnection(
|        "\"fullName\"	\"shortName\"	\"espre\"	\"IFNName\"	\"color\"
| \"fagus\"	\"Fagus sylvatica\"	\"fagus\"	\"4\"	\"fagus_sylvatica\"	\"red\"
| \"quercus\"	\"Quercus robur\"	\"quercus\"	\"3\"	\"quercus_robur\"	\"green\""
|      )
|      res <- utils::read.table(
|        con,
|        header    = TRUE,
|        row.names = 1,
|        sep       = "\t",
|        as.is     = TRUE
|      )
|      close(con)
|      res
|    })
| ## Species\ names\ and\ espre\ codes\ of\ species\ used\ by\ Vallet:1 ends here
| 
| ## Local Variables:
| ## org-babel-tangled-file: t
| ## buffer-read-only: t
| ## eval:: (auto-revert-mode)
| ## End:
`----

** evaluation
data and code is *never* evaluated, i.e. 
:PROPERTIES:
:header-args+: :eval never
:END:

The code and data is loaded as follow:
#+begin_src R 
library(devtools)
load_all("./")
#+end_src

and there you go.
 
To debug the code, I enable developer mode in ESS
and add the package =the_name_of_the_directory_./= to be debugged.

* Using the code
I have a second subtree (except of the one containing the code and the
data) which is evaluated (as some lengthy analysis are involved, not on export but before) and contains the actual analysis. 

* Advantages
- All code and data is enclosed in a "package" environment and,
- when the project is finished, it can be easily packed in a package
  and handed on to other users.
- documentation to the functions can be added using roxygen which can
  easily be translated into .rd files


* Final words
I hope this clarifies my approach.

For me, this is a very "natural" approach, as I like to use packages
and capsulate everything in environments / packages, and it seems very
"clean" to me to define the core of the analysis as functions in
=./R/*.R= files, the variables in =./data/*= files and the actual
analysis as mainly a few function calls.

If you have any further questions, please go on and ask.

Cheers,

Rainer





>
> Online Thu, Oct 9, 2014 at 7:37 AM, Rainer M Krug <Rainer@krugs.de> wrote:
>>
>> I'd like to ping this as I think it
>> is an important bug. It e.g. make debugging of R packages, whose source
>> is in org, more difficult, (the .R files are tangled into ./R/) as ess
>> does not find the source of the R code in R as it uses the links in the
>> comments to find the org file.
>>
>> Temporary solution: I created a link to the org file in the ./R directory.
>>
>> Cheers,
>>
>> Rainer
>>
>> Rainer M Krug <Rainer@krugs.de> writes:
>>
>>> When in the tangling taerget a directory is specified, this is not
>>> reflected in the comments in the tangled file when comments are set to
>>> include links:
>>>
>>> ,----
>>> |    * `link' The code block is wrapped in comments which contain
>>> |      pointers back to the original Org file from which the code was
>>> |      tangled.
>>> `----
>>>
>>> In the org file =prodMixStands.org=:
>>>
>>> ,----
>>> | :PROPERTIES:
>>> | :header-args+: :tangle ./R/update.cache.R
>>> | :END:
>>> | #+begin_src R
>>> | test
>>> | #+end_src
>>> `----
>>>
>>> results in the file =update.cache.R= in a link to the file
>>> =prodMixStands.org= and not =./../prodMixstands.org=:
>>>
>>> ,----
>>> |  ## [[file:prodMixStands.org::*Begin][Begin:1]]
>>> `----
>>>
>>> This is particularly bad when using ess and developer mode to develop a
>>> package.
>>>
>>> Cheers,
>>>
>>> Rainer
>>
>> --
>> Rainer M. Krug
>> email: Rainer<at>krugs<dot>de
>> PGP: 0x0F52F982
>>

-- 
Rainer M. Krug
email: Rainer<at>krugs<dot>de
PGP: 0x0F52F982

[-- Attachment #2: Type: application/pgp-signature, Size: 494 bytes --]