← Changelog index

---

v1.4.4 — Revert KOMA-Script scrbook back to plain book

KOMA-Script scrbook (adopted to chase a Springer-Monographs-in-Mathematics look) caused more churn than it was worth: it conflicts with titlesec (had to be replaced with KOMA’s own heading commands), its DIV/trim tuning had to be re-derived at every page-size change, and each of those changes reopened “does every table/code listing still fit” from scratch. Reverted preamble.tex to plain book, keeping the typographic improvements from the previous entry via plain-LaTeX equivalents instead of KOMA options:

• \documentclass[12pt,twoside]{book} + \usepackage[margin=1in, bindingoffset=0.3in]{geometry} in place of scrbook’s fontsize=/ DIV=/paper= options.
• \setlength{\parskip}{0.5\baselineskip} (keeps \parindent, unlike the parskip package) in place of KOMA’s parskip=half.
• \linespread{1.15} carried over unchanged.
• Chapter/section styling reverted to the original titlesec-based \titleformat rules (the same visual result: a colored rule above/below the chapter title, no numeric chapter label, colored section headings) instead of KOMA’s \RedeclareSectionCommand/\chapterlinesformat.
• The unnumbered-front-matter-chapter mechanism (unnumber_chapter() in build_latex.py, \setcounter{chapter}{-1} before Chapter 0) and the table column-ratio/\small fix from the previous entries are class-independent and needed no changes.

235 pages on A4, compiled clean: zero errors, zero Overfull \hbox, zero Missing character warnings.

v1.4.4 — Typography pass: bigger font, more paragraph spacing

Follow-up to the KOMA-Script switch: the default 11pt/dense-paragraph look read too small and cramped.

• Bumped to fontsize=13pt, parskip=half (a visible gap between paragraphs, not just first-line indent), and \linespread{1.15}.
• Switched DIV from a hand-picked value to DIV=calc, letting KOMA’s typearea compute the type area that actually matches the chosen font size — a fixed DIV was triggering KOMA’s own “no optimal type area settings” warning and produced a narrower-than-intended text column.
• A smaller physical trim size (first A5, then Springer’s actual 155mm x 235mm monograph trim) was tried, on the theory that a fixed-layout PDF page shrinks as a whole to fit a phone/tablet screen, so a smaller page-to-margin ratio reads larger once shrunk. Reverted back to standard A4 — every trim change reopened the “does every table/code listing still fit” question from scratch, which wasn’t worth it here.
• Along the way, fixed two things a narrower page made newly visible:
  • Three Lean/Python code blocks used manual leading-space alignment to visually continue a trailing comment onto a second line (a trick that only holds up on a wide page); with breaklines=true on a narrower measure, each wrapped line’s leading spaces compounded into a progressively cascading rightward drift. Rewrote each as a single plain trailing comment that wraps normally instead.
  • The two notation-reference tables’ four columns were split evenly (25% each), which was fine on a wide page but made the long “Meaning” column wrap onto far more lines than necessary once the page narrowed. simplify_tables() now gives 4-column tables an asymmetric 34/20/28/18 split, and wraps every table in \small — a real page-spanning longtable turned out to be broken in this toolchain’s booktabs+longtable combination even in total isolation (confirmed with a minimal, unrelated test file), so every table in this book remains a non-breaking block, making its per-page fit sensitive to exactly this kind of column-width tuning.