BOOKS: Advanced UNIX(TM) Programming

[John Gilmore]

In article <737 at drutx.UUCP>, mrl at drutx.UUCP (LongoMR) writes:
    (The double indent is me, gnu, speaking:)
> > Laura and I looked at this book at a friend's house and our opinion
> > was that if you only want to write for System V, it might be OK.
> > If you care about the larger Unix world, pass it up.  He seemed to
> > have kind of the right attitude, but all the little details are
> > wrong, and he totally ignores the non-att world.
> 
> Wow, that seems like a pretty definate statement to be offered without
> examples, wouldn't you say? I would be interested to know not only
> your opinion, but how you arrived at it. Note that this is not a flame.
> I just bought the book and if I am going to be wasting my time, I'd rather
> not read it. I would, however, like to know how you came to your conclusion.

As I said, we only looked at the book at someone else's house.  We opened
it in a few places and found enough odd things.  He suggests that you read
directories by opening them and doing 16-byte reads into a struct; that's
somebody who's never heard of the fast file system or portability if I've
ever seen it.  Laura said at the time that his preface statements about
the history of Unix were wrong -- not in major ways, but in small ways.
One place I opened it he used a strtok() function which was the first
I'd heard of it -- though later I noticed a PD implementation of it
in my net.sources archives, it didn't appear in BSD systems or V7 or 32V.

I'm not saying it's a bad book, I'm just saying it's too focused on SV
for my taste.  I care more about portability than that.  I would undoubtedly
learn things from it, and I know enough about non-Sys V systems to know
when what he's saying is unportable or wrong -- but how much of his
audience is in that position?