Software Engineering Radio - the podcast for professional software developers SE Radio 695: Dave Thomas on Building eBooks Infrastructure
Nov 19, 2025
Dave Thomas, co-founder of the Pragmatic Bookshelf and author of The Pragmatic Programmer, talks about building eBook infrastructure and tooling. He covers what eBooks and EPUBs are, the shift from MOBI and PDFs to reflowable formats, why plain-text and Markdown won out, pipeline architecture and preprocessing, conversion to EPUB/PDF, CI builds, handling code and cross-references, and the role of humans plus AI in publishing.
AI Snips
Chapters
Books
Transcript
Episode notes
EPUB Is Essentially Zipped HTML
- EPUB is the de facto e-book format and is just a zipped collection of HTML, assets, and manifests that readers unzip and render with a browser-like engine.
- Dave Thomas explains you can unzip an EPUB to inspect XHTML and navigation files, which is why EPUB converged as the cross-platform standard.
Pragmatic Books Used Images For Code On Early Kindles
- Early Kindles required MOBI and lacked features like preformatted code, so Pragmatic Bookshelf embedded code as images to force correct rendering.
- Dave recalls creating image-based code listings because Mobi couldn’t reliably support pre tags, causing painful zoom and pagination issues.
Use Reflowable EPUBs Not Fixed PDFs
- Prefer reflowable formats (EPUB) over fixed PDFs for readable electronic books because reflow lets readers change font and window size without breaking layout.
- Dave Thomas warns PDFs are page-based and hard to reflow accurately because text is stored in tiny positioned chunks.
