Re: The type of argv in K&R2

[Ben Bacarisse]

candide <ca*****@free.invalidwrites:

K&R2 §5.10 tells that the second parameter to the main function,
usually called argv, has type pointer to character strings.

Some quotes from the book :

-------------------- 8< ----------------------------------------
"the second (argv, for argument vector) is a pointer to an array of
character strings"

"Since argv is a pointer to an array of pointers,"

"Since argv is a pointer to the beginning of the array of argument strings"
-------------------- >8 ----------------------------------------

In fact, argv is rather an array of strings, as explained in the
preceding § :

argv can't be an array. It is a function parameter and arrays can't
be passed to functions in C. There is indeed an array "out there" but
all main can ever see is a pointer to it.

Not everyone likes the term "array of strings" -- it can be see as a
bit woolly. What is not in question (I hope) is that, inside main,
argv is not an array: it is a pointer to the first element of an array
whose elements are character pointers. When main is called, a further
guarantee is made: that these pointers (if they are not NULL) point to
strings (i.e. that the character arrays pointed to are
null-terminated).

You will hear people say "argv is an array of strings" and that is fine
at the bus stop (if you frequent that sort of bus stop) but only
because everyone know what it really meant: that the second parameter
passed to main is a pointer to an array of pointers that point to strings.

--
Ben.

[Richard Heathfield]

candide said:

Ben Bacarisse a ecrit :

>argv can't be an array.

In K&R, the argv parameter is declared as :

char *argv[]
so the type of argv is array to pointer to char.

No, it's a pointer to pointer to char. It's just that it points to the
first element in an array of pointers to char.

And the C90 standard agrees :

----------------------- 8< ---------------------------------
5.1.2.2.1 Program startup

(...)the array members argv[0] through argv[argc-1] (...)

That's fine - those pointers /are/ array members.

(...)to by the argv array shall be (...)

That isn't fine - it's sloppy wording. But we've been round this loop
before, and the committee aren't about to change anything.

K&R declares argv to be "a pointer to an array of pointers" and it seems
to me to be incorrect.

Sloppy wording again. But note that, just after your cite, K&R says very
unsloppily that "argv... is a pointer to pointer to char", which is right.
It /also/ says "pointer to the beginning of the array of argument
strings". If we take "beginning of the array" as meaning "first member of
the array", that's actually right too.

But yes, what you've succeeded in noting is that both the Standard and K&R2
are guilty of describing argv in rather woolly terms. It has been noted
before, and no doubt it will be noted again.

--
Richard Heathfield <http://www.cpax.org.uk>
Email: -http://www. +rjh@
Google users: <http://www.cpax.org.uk/prg/writings/googly.php>
"Usenet is a strange place" - dmr 29 July 1999

[Keith Thompson]

Richard Heathfield <rj*@see.sig.invalidwrites:

candide said:

[...]

(...)to by the argv array shall be (...)

That isn't fine - it's sloppy wording. But we've been round this loop
before, and the committee aren't about to change anything.

[...]

I agree that it's sloppy, but a friendly reading (one that makes it
correct) is possible.

If you interpret the phrase "the argv array" to mean "the array to
whose first element argv points", then the wording is correct. Note
that this does *not* imply that argv is itself an array.

But it's perfectly reasonable to intepret the phrase "the argv array"
to mean "the array named ``argv''", and there is no such array
(there's only a pointer named ``argv''). So yes, the wording is
sloppy.

--
Keith Thompson (The_Other_Keith) ks***@mib.org <http://www.ghoti.net/~kst>
Nokia
"We must do something. This is something. Therefore, we must do this."
-- Antony Jay and Jonathan Lynn, "Yes Minister"