Unless you're already a badass, sitting down in front of 400 blank pages is too hard.
A lot of people would like to write a novel. Novels are hard. Short stories (that might later became chapters) might be a better starting point.
Nonfiction is a little easier, with less of a step change between short and long form. But, smaller chunks can still be a preferable starting point.
Did your setup ultimately prove useful in writing your book? Or did you give up part way through and that helped you get to the task of actually writing?
Python's main purpose was to enable thousands of cross-references across the books, and to transform a large, every-changing text database into a pleasing output, where different parts of the database had different output formats.
Perhaps a TeX wizard could have done it all in pure TeX, but I never could have. Besides, who wants to learn all the intricacies of TeX/LaTeX when the task in Python is so straightforward? To this day, I can't imagine a different approach.
Joining the Python and TeX pieces together into simple scripts was trivial. The Python code evolved as I got into the books, so I would recommend a rapid-iteration model of code development, rather than building a perfect edifice before starting.
in the end it was sublime (the software) and focus (not a software).
after the manuscript phase it went to google docs (one doc for each chapter) for structure and work together with the editor. then indesign for the triumvirate of me, the designer and the editor.
update: just looked at your https://github.com/poulter7/ipynb-tex, damn that looks familiar... I stand with my I) recommendation
Do you have any other tips and suggestions on how to "stop Wittgensteining"?
obsessing over every word, over every meaning of every word might make the communication 0,5% more accurate, but weakens (to seriously hurts to kills) the overall message. accept that you will be misunderstood, but try not to.
I) just write
first attemp i made a super fancy setup - google chrome book with custom linux installed that automatically transformed markdown via pandoc into a custom styled PDF everytime i saved. was fun to create this setup, but i wasted time not writting the actual book.
II) don't wittgenstein
i wanted to be absolutely correct, so i started with defining word, what do i mean with website, domain, webproperty, webpage, page, link, external duplicate content, internal duplicate content, ... or any other technical term. wittgensteining is yak shaving for authors. don't do it. just write.
III) a manuscript is not a book
after you are finished, the real work start. a manuscript does not have the same feeling, read-feeling, read flow as a real book. get a professional editor, work with him/her. fire him/her if not the right match, find the right match, start again. the right editor is as much a factor for the success of the book as the author.
IV) live with the law of strawberry jam
the further you spread knowledge/jam the thinner it gets. yes, in a 1:1 talk, in a 1:12 people workshop you can communicate more knowledge than you can with a book. talk, interaction, any form of communication with exchange, has direct feedback - something unbelievable important for information communication. a book is a one way street with - if at all - delayed feedback. you can always continue writting, pack in more knowledge, more details, more clarificaiton, more scenarios, but then the book will never finish. at one point you have to stop.
V) cut away
go through your book again, and question every word, every sentence, every paragraph, every chapter. if the word, the sentence doesn't ad substential value, get rid of it. your book will be better. it's not about quantitiy of words, it's about what you want to say.
VI) let go
& publish, just do it.
[1] "Understanding SEO - A Systematic Approach to Search Engine Optimization" https://www.fullstackoptimization.com/b/understanding-seo ISBN: 978-3-200-05426-4