Twenty Years of Pandoc

On August 3, 2006, the creator of Pandoc uploaded its first version to their website, releasing it under the free GPL license. Pandoc 0.1 began with around 3000 lines of Haskell code and relied solely on GHC’s standard library. It was capable of converting Markdown, reStructuredText, HTML, and LaTeX documents into other formats including RTF and S5.

The project was unpredictable, evolving into a project with over two hundred releases over twenty years, becoming the most popular Haskell program and gaining widespread support.

The initial decision to write Pandoc in Haskell was unwavering. Haskell's suitability for document conversion was clear, and the author had heard about Haskell from a friend's blog and an introductory book. Intrigued, they read A Gentle Introduction to Haskell, but to truly learn a language, they used it. The approach of parsing Markdown with parser combinators and generating an abstract syntax tree (AST) proved more reliable and extensible than previous regex-based implementations.

This design choice enabled support for multiple conversions (N × M) and led to the addition of reStructuredText readers and LaTeX writers.

As word spread, the project gained visibility through collaborations, such as with a Turkish developer who helped package pandoc for Debian Linux. Pandoc 0.3 introduced the DocBook writer and standardized Markdown syntax for footnotes. Version 0.4 expanded functionality with markdown tables, definition lists, and enhanced list support, among other features. The project's move to the Hackage repository in 2007 further increased accessibility.

The release of Pandoc 1.0 in 2008 added writers for MediaWiki, GNU Texinfo, OpenDocument, and ODT, alongside syntax highlighting capabilities using the highlighting-kate library. Pandoc's growth was guided by its utility for the author's academic work and its increasingly diverse set of supported formats.