Re: Fwd: ./configure failure, .log attached

[Guy Harris <guy () alum mit edu>]

On Jan 26, 2011, at 3:09 PM, Cameron Elliott wrote:

> ./configure told me to send my ./configure failure to the list.
> (because it failed)
> I hope that is correct.

Yes.

> If you understand why it is failing, please explain to me.

It's failing because one of the checks it's doing is assuming that an OS with IPv6 support has a <netinet6/in6.h> 
header file, but your version of Ubuntu - and possibly other Linux distributions - don't.

The Single UNIX Specification seems to imply that you include <netinet/in.h> to get address etc. definitions for IPv6; 
it doesn't mention <netinet6/in6.h>.  The "Advanced Sockets API for IPv6" RFCs (RFC 2292 and RFC 3542) don't mention 
<netinet6/in6.h>, either - they mention <netinet/in.h> and <netinet6/ip6.h>.  Thus, I don't think you could argue that 
the lack of <netinet6/in6.h> is a bug; the configure script should be changed not to assume it's there on IPv6-capable 
systems.

I don't see anything obvious in tcpdump that *cares* whether <resolv.h> defines res_state_ext or whether it has an 
"nsort" member, so, unless somebody can give a good reason why the "checking for res_state_ext" or "checking for nsort 
in res_state" checks need to be done, I will remove them.

> Also, the email address in ./configure should be updated, it is wrong, it says:
> configure:7867: result: no
> configure:7874: error: Report this to tcpdump-workers () tcpdump org, and
> include the config.log file in your report.

I've checked a fix for that into the main and 4.1 branches.-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.