Frequently Asked Questions file will not exist

[Don Libes]

>In article <1990May5.070317.11503 at agate.berkeley.edu> raymond at math.berkeley.edu writes:
>>The Frequently Asked Questions file project has died due to lack of interest.

>Which is to me a very strong signal that the MS-DOS (and UNIX) questions
>aren't a small, fixed set that can be answered in a short FAQ.

Many of the questions are better addressed in the literature already.
If your answer is so breathtakingly masterful or comprehensive, then
you should publish it so we can refer to it in the future.

I encourage people to answer questions by citing references to easily
accessible sources.  For example:

	Q: How does malloc work?
	A: See K&R2, page ...

	Q: Can we add "complex" to C?
	A: See JCLT, vol 1, page ...

	Q: How do I do reverse video in C?

Questions like the last one are often underspecified and cannot be
answered immediately.  (Such questions indicate people are in over
their head, and couldn't really use the answer if you gave it to them
anyway.)

I learned a long time ago that giving people direct answers guarantees
they will come back to you for the next question.  Instead I show them
how to answer questions for themselves.

"Don, how do you make a pointer to a function returning an int?"

I say, "How do I make a pointer to a function returning an int?  Well,
gee... I dunno offhand, but I think "C: A Reference Manual" has a
chapter on declarations.  Hmm, table of contents say "chapter 4 ...
declarations" aha!  [Turn to chapter 4] Ohhh there is a nice
discussion of this very topic called "composition of declarators" on
page 71.  I wish I had time to read it, but ... well why don't you go
ahead and read it for yourself."  (I say this very sincerely.)

Most of the time I send people to K&R, H&S, the Standard, (half a
dozen other books I've got on my desk,) USENIX or other proceedings,
and various periodicals such as BSTJ and the C Users Journal, always
explaining why I selected the resource I did.  Extremely rarely I
suggest asking comp.lang.c.  Once or twice I've suggested mailing to
someone like Doug or Chris.  I'm rather pleased that I've never been
embarrassed reading a posting to comp.lang.c from our site.

My answers often end with "you really should have a [copy, subscription,
whatever] of that if you're going to be referring to it often."

Virtually the only questions I get anymore are really difficult ones
that aren't written down anywhere, and an occasional question from a
new user who hasn't met me before.

The point of all this is that the FAQ is in a handful of books and
periodicals.  Of course you can't package that into a short essay.
The only thing you can package is a bunch of references.  (If I wrote
the FAQ, it would just be references.)  Even ANSI C already has a
large number of books and articles written about it.

To those of you who can't afford these books or subscriptions I say:
You (or your computer center) bought a $10,000 machine without the
$500 set of readings!?!?  Evidentally you priced an incomplete system
and could never have afforded the complete system.

To those of you who can't make head or tails of the books, find a
local guru, before blindly posting to the world.


To summarize: If you want to make an FAQ, it should be primarily
references.  If it really answered all the common questions that we
see here, it would be several books.  And they exist already.
(Believe me, there are very few things that haven't been published!)

I really am serious about one-line answers.  I've seen only one person
doing this.  Cite the work, and then resist the temptation to
paraphrase it, or comment on it.

I realize there are valid articles to be posted to this newsgroup.
Let's eradicate the junk so we can see them.

Don Libes          libes at cme.nist.gov      ...!uunet!cme-durer!libes