A review article on conformal field theory 'cft33.tex' (145 pages), with also a much briefer version 'minimal.tex' (24+13 pages) and even briefer versions 'crash.tex' (12 pages) and 'crash2.tex' (11 pages).
Why would a review article on theoretical physics end up on GitHub, and not just on Arxiv? The idea is that review articles would be better if they were collaboratively written. An interesting experiment was done with the review of AdS/CFT integrability, where dozens of authors each contributed a relatively small piece. Here we test another approach: to open an initially single-authored text to collaboration by anyone interested.
To have the article on GitHub could already be useful by enabling readers to directly correct misprints, without having to go through the author. The hope is however that more substantial contributions will be made. (If this occurs, it may become necessary to modify the authorship of the resulting text.)
For some indications on how to compile this article, see the comments in the Latex source file cft33.tex. The necessary Latex packages are all standard except maybe underbracket.sty, which had to be included in the source files for Arxiv.
The BIB file contains hundreds of references, many more than are really used. The superfluous references have not been removed, because some of them could be useful later, in particular for versions that would include guides to the literature.