Reproducible Research with Org-babel

org-babel is a set of features on top of org-mode

org-mode summary

org is for organise! * for levels as # is for markdown. use multiple * for subheadings (e.g. ** etc)

• You can create tables M-x org-table-create-or-convert-from-region (?) You can use it as a spreadsheet, with rows and columns operations.

• headlines can have properties key-values pairs (C-c C-x p)

• org also as todo lists (C-c C-t), with schedule (C-c C-s) and deadlines (C-c C-d) that you can see in the agenda. (C-c a a)

• Org-mode is incredibly broad, from task management to writing literate code and papers. The basic principle is that it has functionality related to hierarchical organisation of text (under headings).

• All documents can be exported to all sorts of outputs by calling org-export. You can export to all sorts of html, pdf or other markup langages (Markdown, latex etc).

• code blocks for any language:

    #+begin_src python
    import numpy
    #+end_src
  

  (You can use <s tab and it will create the source block) C-c ' will open another buffer with the major mode of the language used in that language (as if visiting a python file); C-c ' will take you back to the org file.

  C-c C-c will run that code block, and it shows the result below.

• When the output is a list of string, for example the “shell” example is showing the output as a table of files. Same type of table as the one shown before.

Example workflow

• We’ve starting writing an org file, available in this git repo.
• We start with a C++ code which generates some data for us, in this case random 1D browninan motion.
• We want to put this small C++ snippet inside a code block, see file for examples.
• C-c C-c will write the stdout of the
• we can name the code block with: #+NAME: initial-data above the code block
• In a following code block, we can refer to the output to the above python :var timeseries=initial-data and a variable timeseries will be read instead.
• print won’t show the output of the code, but you can use :results output.
• Results/Variables are not cached, so it will ask you to run the dependants code blocks.

• exports won’t show the output, unless :exports both

• Then we can have inline results within the text: src_python[:var timeseries=initial_data]{} the code will be inserted within the {}

• to simplify things we can create a code block that then we can call, we have to name the code block, and then use it as a function with: #+CALL: mean(initial_data) where mean is the name of the block.

• Inline calls can be used with call_mean(input_data) inside the text. Then use C-c C-c to evaluate the function as an inline codeblock

• you can include functions from separate files using #INCLUDE:

• A :prologue header argument can pre-pend code to a code block. For example: :prologue "import numpy". You can add this at all kind of levels. On the code block #+begin_src line, or in a property drawer at the document level or any subsection. You can also restrict headers to one language.

• you can also use it to show figures.

• tangle: can be used for literate programming.
• emacs configuration as an org-mode file.
  • org-babel-tangle extracts all the code from an org mode that then it can be used (e.g., by emacs to use to load the configuration file).
  • (org-babel-load-file "file") it’s a org-bable helper function that will untangle the org-mode file when load.