Do we need a "Stevens" book?

[Olwe Melwasul <hercynianforest@gmail.com>]

I've not gotten very far with this idea; no one seems interested, but
I'll try it here anyway...

It seems to me that Emacs needs a W. Richard Stevens-style book. As
you may know, Stevens wrote the "Advanced Programming in the UNIX(R)
Environment" textbook that many of us used in college. Or maybe Emacs
needs something along the lines of the many "Linux gnarly/wooly
internals" books. Anyway, I would love to see a book that got into the
nitty-gritty of Emacs/elisp -- just like you see discussed here every
day on the help-gnu-emacs list.

Here's an example: comint. How do you effectively use comint? When
should you use comint? Okay, I can Google around and find one-off blog
discussions here and there about comint; I can read them all; I can
get confused; I can kludge something together ... and then find out
later that what I've done (as well as bloggers A, B, and C) is really
not "best practice" use of comint, i.e., that how I've used comint is
overkill or could have been done much simpler with <some other>.el.
Wouldn't it be nice to have one go-to source/book that thrashed out
comint usage once and for all?

Just skimming through all the elisp material (books, Internet, etc.),
it seems like a hodge-podge on a continuum between gems and junk just
waiting for a clear-speaking Richard Stevens to whip it all into
shape. Sure, the "official" texts will get you pretty far, but no way
are you ready to be a "best-practices" guru. The printed books seem
more like a "cookbook" than a real Stevens-style book. Maybe I'm all
wrong, but I think I like what the Racket/PLT people are doing. They
seem to be whipping the Scheme hodge-podge into a decent
best-practices, best-tools order.

Personally I've been admiring Emacs from afar for quite some time. I'm
really an Emacs/elisp newbie, but I've got a writing/technical writing
background. If what I'm saying strikes a chord, maybe I could be a
receiver/collector of a "best-practices-slash-wooly internals" sorta
book project. It would be a free/GNU sorta thing of course ... and
please don't say "I don't think there'd be enough interest in it."

Olwe