From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.1b1; Mac_PowerPC)

Description of problem:
There is a bug in the timeout mechanism of netcat under Linux (and 
IRIX) due to the change in the way setjmp/longjmp is implemented in 
the c library.  


How reproducible:
Always

Steps to Reproduce:
1.pick an IP address that you cannot reach (e.g. behind a firewall, 
non-routable, whatever) call it $host
2.nc -vvw2 $host 1-20
3.notice how only the FIRST attempt actualy times out in the 2 
seconds you specify, and the rest wait for a full tcp timeout?  This is 
broken.
	

Actual Results:  The first connect() attempt is correctly cut short after 
2 seconds, while the rest wait for a full timeout.

Expected Results:  Each attempt should have timed out after 2 
seconds.

Additional info:

This is due to a change in glibc, which decided to go one way, while 
all the other libcs went another- simply because POSIX doesn't care 
one way or the other.

It is broken under every version/distrobution of linux which uses glibc-
2.x.

I have submitted this to debian already, and now you.