Python books

[Andrew M. Kuchling]

evil Japh writes:
>Python" is more a book to be read through, rather than a reference; I'm
>not saying I expected it to be a reference, but that I would like a book
>that was a reference.

	I think the existing free documentation at
http://www.python.org/doc/ is fine for reference purposes, and
multiple reference guides aren't needed.  Reference docs have the most
complete and detailed coverage, and are subject to the most nitpicking
flaws, so the division of effort resulting from two different sets of
reference docs costs you a lot of accuracy and completeness.  We
should concentrate on improving them.

	For an example of why this is bad, look at some of the
O'Reilly Java books, such as the Swing book.  Some of the material
essentially duplicates the Swing API docs, but you can't trust that
material because, given the lead time required for book publication,
some details may have changed, so you have to look at Sun's docs for
definitive information anyway; having another reference-like work
doesn't help anything, because it might be wrong.  The reference-like
material also makes the Swing book painfully dull to read, and
probably soul-destroyingly boring to write, too.  Focusing on tutorial 
aspects would have made the book smaller, less likely to get out of
date, and more readable.

	Multiple *tutorials* and other learning documentation, on the
other hand, are definitely good, since different people like different
presentations.

-- 
A.M. Kuchling			http://starship.python.net/crew/amk/
    "Where does this one come from? Have you been raiding poor Holinshed
again? Or does Plutarch bear the brunt of your depredations?"
    "Bits of things, here and there, but it's mostly mine, for once."
    -- Ben Jonson and William Shakespeare, in SANDMAN #75, "The Tempest"