holding the conn->data value

responded with a single status line and no headers nor body. Starting now, a
HTTP response on a persistent connection (i.e not set to be closed after the
response has been taken care of) must have Content-Length or chunked
encoding set, or libcurl will simply assume that there is no body.

To my horror I learned that we had no less than 57(!) test cases that did bad
HTTP responses like this, and even the test http server (sws) responded badly
when queried by the test system if it is the test system. So although the
actual fix for the problem was tiny, going through all the newly failing test
cases got really painful and boring.

as appropriate for platforms that don't have autotools support

struct, and there are domain/search entries in /etc/resolv.conf, the domains
of the options struct will be overridden.

defining HAVE_SIGNAL_H if the header is available.

Added a check in configure that tests if the sig_atomic_t type is
available, defining HAVE_SIG_ATOMIC_T if it is available. Providing
a suitable default in setup_once.h if not available.

Added a check in configure that tests if the sig_atomic_t type is
already defined as volatile, defining HAVE_SIG_ATOMIC_T_VOLATILE
if it is available and already defined as volatile.

This is done to back out changes done in revisions 1.77 and 1.75

This is done to back out changes done from revisions 1.213 to 1.217

This is done to back out changes done from revisions 1.6 to 1.10

have proper TERM and INT signal handlers implemented.

servers hash, and adjust message arguments accordingly.

be clear of warnings. Uncomment it if this module is further modified.

The "warnings" module requires perl 5.006 or later. Previous perl
versions don't have it and die on missing modules.

If the list has only one item avoid sort subroutine.

server holds not only its two main pids, but also the pidfile of the test
server and the 'slavepidfiles' for ftp* servers. This allows a better control
when stopping servers.

Now from runtests.pl when test servers are stopped they are signalled in
sequence TERM, INT and KILL allowing time in between for them to die. This
will give us a chance of gracefully stopping test servers, which we didn't
have when we were killing them in first instance.

libssh2 startup.

out a stack overwrite (and the corresponding fix) on 64bit Windows when
dealing with HTTP chunked encoding.

They all now return ssize_t to Curl_write().

Unfortunately, Curl_read() is in a sorrier state but it too would benefit from
a similar cleanup.

Versions, not to ./Versions and indentation improvments

2006. It turned out we wrongly assumed that the connection cache was present
when tearing down a connection.

multi interface, but I could also repeat it doing multiple sequential ones
with the easy interface. Using Ciprian's test case, I could fix it.

Add failure checking for servers when fork()ed.

Use same code path in 'stopserver' when called with a single or multiple pids.

CURLOPT_VERBOSE set to non-zero, you still got a few debug messages from the
SSL handshake. This is now stopped.