Scanned PDFs, now with tables of contents

@anrddh updated the scanned PDFs, adding comprehensive tables of contents:

• 4th edition, 1998
  • higher resolution, better OCR, smaller file thanks to better compression
  • previous release
• 6th edition, 2001
  • now including metadata, like title and author
  • previous release

For more details, see #166

Alpha drafts of ebooks from markdown

This is a rough draft! Issues are expected.

what's here

• epub - made by pandoc from markdown
• pdf - made by calibre from the epub version

features

• we can proofread and annotate on offline devices! 🎉
• includes the WIP effort to turn images to markdown tables
• includes the redrawn diagrams, as vector graphics

known issues

• chapter links in the Preface chapter don't work
  • there are many more chapter links that have never been fixed up to work at all
• some tables have irregular column widths
• the TOC numbering is slightly out of sync, counting the frontmatter as chapter 1, About as chapter 2, Preface as chapter 3, Chapter 1 as chapter 4, etc.

epub

• the epub has the svg diagrams and formulae, not the png fallbacks
• Apple Books isn't showing code blocks, on desktop?

pdf

• an extra page break before section 12.1
• Bluefire appears to have some intermittent font glitches with italics
• some lines are split between pages, and some lines are truncated at the right

what's next

• pandoc worked for making a legible first draft, but using something else probably would help or sidestep the issues here. so, exploring alternatives is a good idea.
• these were made with default layouts for the software; customizing font sizes, margins, and the like should maybe wait

A better scan

About this copy

This is a scanned copy of the 4th printing, 1998. It's shared for reading, and for improving the Markdown copy in our Github repo.

How it was made

@pronoiac had the spine / binding removed and fed the pages through a scanner. Steps and software used:

• scanner gave 600dpi grayscale, as 3.6 gigabytes of png files
• Scantailor Advanced (in Docker): deskew the pages and render the pages as 300dpi black and white (1-bit) tiffs - 30 megabytes
• tiff2pdf and pdfunite: turn those many tiffs into one pdf
• OCRmyPDF: OCR with Tesseract, add title and author to the pdf, apply lossless JBIG2 compression - 24 megabytes

Other notes

• It’s higher resolution, though an older printing (4th printing, 1998) than the previous scan (6th printing, 2001).
• OCR is better than the previous scan - searching for keywords or phrases usually works
• why not the grayscale PNGs: space constraints on Github releases, and dubious value for space
• ebooks from the Markdown version are getting closer
• see #137 for some of the thoughts behind this release

Cleaned up Safari ebook

In July 2018, @eshellman worked on the epub version converted from Safari, fixing the cover and some html errors, and adding a .mobi version. This isn't generated from the markdown files, which should be cleaner.

Original sources

These are sources we're working from.

The original scanned PDF:
This is a copy of the original scan in one file. Readers can get one PDF instead of two, and the repo could be a little easier to work with.

An EPUB converted from O'Reilly Safari Books Online:
It's a much cleaner source to work from.

(The tar.gz and zip files are unintentional.)