On Wed, Dec 11, 2013 at 06:49:17AM -0500, Robert P. J. Day wrote:
> a general question for the masses -- what value do you see in
> linux books these days?
I'll assume you mean Linux and common userland, and not the kernel
itself.
> if people on this list still buy books, how do you judge whether
> something's worth purchasing? what do you look for?
* Technology or trend du jour-agnostic. Books about learning Python 2.x
are to be avoided since they'll be obsolete soon. Ditto for books on
the latest web framework, desktop, distro, etc. In other words…
* Focused on concepts and so on rather than implementations of these
concepts, though case studies thereof are acceptable provided that the
learning exercise doesn't require too much use of said implementation.
E.g., a book about programming language design that touches on BASIC
is fine as long as I don't lose much of the worth of the book if I
can't or don't want to play [1] with BASIC. Ditto for a book on
cryptography that compares and contrasts TLS and SSH.
* If implementation-focused:
* Must be about technology that will stand the test of time. E.g.,
a study of the engineering of various computer architectures, or
(to some extent), the design of one of the original influential
Unices.
* Either:
* Must be a compendium of scattered knowledge that is otherwise
a bother to aggregate and filter, or must provide a not
readily available angle on the matter.
* Must be to provide an update on best current practice that
would be otherwise hard to gleam from readily available but
still widely referenced documentation, and must provide an
immediate benefit. Had a certain Tkinter-focused site not
existed, a *small* book on modern Tk programming would have
been acceptable. The same could be said of a C++ book [2] on
modern practices, esp. in light of C++ 11 and maybe touching
on C++ 14 and the ideas espoused in some of the proposals.
* Reasonably priced. I don't buy 70$ books unless they transcend the
currently related technology and are packed. Alternatively, if the
book is a reasonable alternative to certain university courses that I
would object to for various reasons (e.g., too much boilerplate, full
of Java crud, etc.), and maybe has a few *interesting* and stimulating
exercises [3], there would be some value, esp. compared to a
few-hundred-dollar course with a name conceived more for marketing
rather than accuracy an descriptiveness and poorly vetted TAs.
* DRM-free, preferrably PDF.
Alternatively, the book could be an interesting story: interviews with
the designers of some interesting programming languages, thoughts from
some **exceptional** programmers on code they particularly liked, etc.
> […]
>
> * available in e-form
[1]: Quite the case here.
[2]: I'm not a C++ programmer. The assumption in this context would be
for somebody like a student that just learned incidentally as part
of their university or college education, esp. if from some likely
outdated reference matter (big, slow-to-publish, overpriced large
textbooks).
[3]: Like certain books on purely functional programming, though half
the challenge was the language itself, not the purely functional
programming.
Regards,
Alex Pilon
Attachment:
pgpFbnK954mfy.pgp
Description: PGP signature