Compile the 'sock' program in the book 'TCP/IP Illustrated Vol.1'

hi,

i am running Linux and want to compile and use the 'sock' program comes with
Stevens's book 'TCP/IP Illustrated Vol.1'. but the code simply can't pass the
compiler with following error,

,----
| bash-3.00$ make
| gcc -ansi -Wall -Dsun -D__STDC__=0 -c -o loop.o loop.c
| <command line>:5:1: warning: "__STDC__" redefined
| loop.c: In function `loop':
| loop.c:117: error: `caddr_t' undeclared (first use in this function)
| loop.c:117: error: (Each undeclared identifier is reported only once
| loop.c:117: error: for each function it appears in.)
| loop.c:127: error: parse error before numeric constant
| make: *** [loop.o] Error 1
`----

in fact, in the Linux system, there is a sys/types.h which defined the caddr_t
type and this file seems have been already included in the source code, so i
don't know why the type still can not be recognized.

had anyone here compiled the code with success? or, is there any other similar
tool freely available ?

thanks.

--
steven woody (id: narke)

Jesse: You want to know why I wrote that stupid book?
Celine: Why?
Jesse: So that you might come to a reading in Paris and I could walk
up to you and ask, "Where the fuck were you?"
Celine: [laughing] No - you thought I'd be here today?
Jesse: I'm serious. I think I wrote it, in a way, to try to find you.
Celine: Okay, that's - I know that's not true, but that's sweet of you
to say.
Jesse: I think it is true.

- Before Sunset (2004)

In article <(E-Mail Removed)>, Steven Woody <anti-(E-Mail Removed)-post-to> writes:
>
> i am running Linux and want to compile and use the 'sock' program comes with
> Stevens's book 'TCP/IP Illustrated Vol.1'. but the code simply can't pass the
> compiler with following error,
>
> | bash-3.00$ make
> | gcc -ansi -Wall -Dsun -D__STDC__=0 -c -o loop.o loop.c
> | <command line>:5:1: warning: "__STDC__" redefined

Why are you defining __STDC__ at all, much less to 0? __STDC__
should always be defined by the implementation. If this is from the
makefile that you got with the source for sock, be aware that _TCP/IP
Illustrated_ v1, while an excellent book, is more than ten years old,
and the makefile might require some reworking for newer systems.
(Though even in 1994 defining __STDC__ explicitly was almost certain-
ly wrong.)

> | loop.c: In function `loop':
> | loop.c:117: error: `caddr_t' undeclared (first use in this function)
> ...
>
> in fact, in the Linux system, there is a sys/types.h which defined the caddr_t
> type and this file seems have been already included in the source code, so i
> don't know why the type still can not be recognized.

Because it hasn't been defined. Including sys/types.h very likely
doesn't guarantee that a particular typedef in it has been processed,
because in most implementations sys/types.h is full of conditional-
inclusion directives (#if and friends).

In the Linux sys/types.h I'm looking at now, caddr_t is only defined
if the macro __USE_BSD is defined. Note that this is an identifier
reserved to the implementation, so you shouldn't be defining it
(unless instructed to by the implementation's documentation - and
that would be a poor choice by the implementor).

As it turns out, __USE_BSD is defined for you (in features.h, which
is automatically included by a raft of other headers) if the "feature
test macro" _BSD_SOURCE is defined. You define _BSD_SOURCE and you
get caddr_t. The GCC default on Linux (and, I think, most Unix
platforms, using the stock GCC build) is to define _BSD_SOURCE unless
you use the -ansi flag. If you do use -ansi (or an equivalent), you
have to define _BSD_SOURCE to get the BSD types, macros, and so on.

You can read more about feature test macros in the Single Unix
Specification, which is available free of charge at [1].

So as a first attempt, I suggest getting rid of that bogus definition
of __STDC__ and substituting a -D_BSD_SOURCE.

> had anyone here compiled the code with success? or, is there any other similar
> tool freely available ?

Stevens mentions some similar tools right in _TCP/IP Illustrated_, in
the appendix that discusses the sock program. One he doesn't mention
(I believe it appeared later) is netcat, which you may already have
as "nc".


1. http://www.opengroup.org/

--
Michael Wojcik (E-Mail Removed)

Today's Carnivore bait: Distracted by the Anthrax song, I let my bin,
laden with goods, crash into a bush.

Michael Wojcik wrote:
> In article <(E-Mail Removed)>, Steven Woody <anti-(E-Mail Removed)-post-to> writes:
> >
> > i am running Linux and want to compile and use the 'sock' program comes with
> > Stevens's book 'TCP/IP Illustrated Vol.1'. but the code simply can't pass the
> > compiler with following error,
> >
> > | bash-3.00$ make
> > | gcc -ansi -Wall -Dsun -D__STDC__=0 -c -o loop.o loop.c
> > | <command line>:5:1: warning: "__STDC__" redefined
>
> Why are you defining __STDC__ at all, much less to 0? __STDC__
> should always be defined by the implementation. If this is from the
> makefile that you got with the source for sock, be aware that _TCP/IP
> Illustrated_ v1, while an excellent book, is more than ten years old,
> and the makefile might require some reworking for newer systems.
> (Though even in 1994 defining __STDC__ explicitly was almost certain-
> ly wrong.)
>
> > | loop.c: In function `loop':
> > | loop.c:117: error: `caddr_t' undeclared (first use in this function)
> > ...
> >
> > in fact, in the Linux system, there is a sys/types.h which defined the caddr_t
> > type and this file seems have been already included in the source code, so i
> > don't know why the type still can not be recognized.
>
> Because it hasn't been defined. Including sys/types.h very likely
> doesn't guarantee that a particular typedef in it has been processed,
> because in most implementations sys/types.h is full of conditional-
> inclusion directives (#if and friends).
>
> In the Linux sys/types.h I'm looking at now, caddr_t is only defined
> if the macro __USE_BSD is defined. Note that this is an identifier
> reserved to the implementation, so you shouldn't be defining it
> (unless instructed to by the implementation's documentation - and
> that would be a poor choice by the implementor).
>
> As it turns out, __USE_BSD is defined for you (in features.h, which
> is automatically included by a raft of other headers) if the "feature
> test macro" _BSD_SOURCE is defined. You define _BSD_SOURCE and you
> get caddr_t. The GCC default on Linux (and, I think, most Unix
> platforms, using the stock GCC build) is to define _BSD_SOURCE unless
> you use the -ansi flag. If you do use -ansi (or an equivalent), you
> have to define _BSD_SOURCE to get the BSD types, macros, and so on.
>
> You can read more about feature test macros in the Single Unix
> Specification, which is available free of charge at [1].
>
> So as a first attempt, I suggest getting rid of that bogus definition
> of __STDC__ and substituting a -D_BSD_SOURCE.

thank you and i will try. but 'nc' you mentioned is enough to me.

>
> > had anyone here compiled the code with success? or, is there any other similar
> > tool freely available ?
>
> Stevens mentions some similar tools right in _TCP/IP Illustrated_, in
> the appendix that discusses the sock program. One he doesn't mention
> (I believe it appeared later) is netcat, which you may already have
> as "nc".
>
>
> 1. http://www.opengroup.org/
>
> --
> Michael Wojcik (E-Mail Removed)
>
> Today's Carnivore bait: Distracted by the Anthrax song, I let my bin,
> laden with goods, crash into a bush.