|
|
Log in / Subscribe / Register

Twenty years of Pandoc

[Posted August 3, 2026 by jzb]

John MacFarlane has published a lengthy retrospective to commemorate twenty years of the Pandoc document converter.

On August 3, 2006, I uploaded the first version of pandoc to my website, releasing it under the free GPL license. Pandoc 0.1 consisted of about 3000 lines of Haskell code, with no dependencies aside from GHC's standard library. It could convert Markdown, reStructuredText, HTML, and LaTeX documents into any of these formats, plus RTF or S5. I had no idea at the time that this would just be the first of over two hundred releases over the next twenty years; that the project would become the most popular program written in Haskell; that I would spend countless hours on bug-fixes, improvement, and project management; that I would collaborate with programmers in many other countries; that pandoc would come to support over fifty document formats; that it would allow automatic generation of citations and bibliographies; that it would become integrated into academic writing tools like Quarto and Jupyter Notebook; that it would be installed on millions of computers around the world.

How did this happen? I want to take advantage of pandoc's birthday to tell the story of the project, as best I can remember it.


to post comments

Philosophy 135

Posted Aug 4, 2026 1:51 UTC (Tue) by wdaly (subscriber, #120437) [Link]

I took Philosophy 135 from John MacFarlane in 2007, and I remember hearing from other undergrads that he wrote Haskell (even though he never mentioned it in class). We were all really impressed by that. Had no idea that pandoc was only about a year old at the time!

beyond markdown

Posted Aug 4, 2026 12:43 UTC (Tue) by dankamongmen (subscriber, #35141) [Link]

The "Beyond Markdown" essay linked from the Pandoc history is pretty good reading.

Pandoc and its relevance in times of LLMs

Posted Aug 7, 2026 15:55 UTC (Fri) by mtthu (subscriber, #123091) [Link]

The retrospective closes with the observation, that an LLM could do the job of translating markup into HTML maybe as well as Pandoc can. While I believe this is true to some extent, I think it comes at a great cost. An LLM consumes orders of magnitudes more energy for the same job as a simple parser would. As energy is a limited resource in most places on this planet, I would argue using Pandoc for translating from one format to the other is also a wise choice in the future.

For fun I asked ChatGPT to estimate the difference in energy consumption for a translation from Markdown to HTML and it gives me a factor of 100 to 10'000 more for its own LLM. Quite a wide range, obviously, but it's shattering even at 100.


Copyright © 2026, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds