Reproducible Research in R with Quarto

Some history and context

• How can we make beautiful documents?

• Specifically, how would a programmer solve this problem?
  
  

• These slides will focus on open standards with open-source implementations.

• The input is always going to be a plain text document, with some special “markup”.

TeX and LaTeX

• 1978: In the beginning, Donald Knuth wanted to write some computer science textbooks, so he created TeX.
• 1980s: System further devloped into LaTeX.
• Write papers and books with beautiful text and mathematics.
• Output is usually PDF. (see XKCD 2304)

Looks like:

\section{Introduction}

In this paper we will consider $x$ and $y$.

\begin{itemize}
\item List item 1.
\item List item 2.
\end{itemize}

\begin{figure}
\includegraphics{graph1}
\caption{An example graph.}
\label{fig:graph1}
\end{figure}

Documents on the World Wide Web

• 1990: Tim Berners-Lee invented the World Wide Web and HTML (HyperText Markup Language).

Looks like:

<h1>Introduction</h1>

<p>In this paper we will consider <i>x</i> and <i>y</i>.</p>

<ul>
   <li>List item 1.</li>
   <li>List item 2.</li>
</ul>

<img src="graph1.png">

Markdown

• HTML is ok, but people wanted more concise formats.
• People wanted documents that looked ok in a plain text editor, and lovely in a web browser.
• A lot of different formats were proposed and used in blogging software, forum software, documentation systems, …
• We have now more or less settled on “Markdown” (2004).

Looks like:

# Introduction

In this paper we will consider *x* and *y*.

- List item 1.
- List item 2.

![Some alt-text](graph1.png)

Literate programming

Donald Knuth had another important idea, literate programming.

(The idea has developed a bit since Knuth introduced it though.)

From a source text file we will weave together a beautiful document:

• Nicely formatted text.
• Code, to precisely document what we did.
• Output from the code such as tables and plots.
  
  

We will also call this knitting or rendering.

Quarto

• Quarto (2022) is the latest iteration of these ideas about creating documents by writing plain text, and literate programming.

• Builds on top of various older packages, so we will see reference to older knitr and rmarkdown packages.

• Allows literate programming in R, Python, other languages.

• Can output many formats, output different formats from same source.

  • HTML document
  • HTML website such as a book
  • PDF (using LaTeX)
  • Word Document

Quarto is developed by Posit (formerly known as RStudio).

Similar ideas: Jupyter Notebooks in Python, Observable in Javascript

Structured data as text

<xml> 
    XML (1990s) is like HTML for data. 
    <item>Easily usable by both people and software!</item>
    <item>People became very excited about XML for a while.</item>
    <item>Turned out to actually be awful for both people and software.</item>
</xml>


{ 
    "JSON" : "JSON (2000s) represents data as you would in Javascript.",
    "items" : [
        "Also very similar to how Python represents data, and lists in R.",
        "JSON was better, but still a bit ugly.",
    ]
}


YAML: "YAML (2001) makes some JSON punctuation optional, uses indentation like Python."
items:
  - YAML has some advanced features you can ignore.
  - YAML is a popular configuration format.
  - YAML is... fine. 
  - Quarto uses YAML for configuration, in the header of documents and in the _quarto.yml configuration file.