hogs Windows machines when libcurl is being used multi-threaded (with > ~50
threads). Andrew Fuller helped us verify and test this.

According to a MSDN web page on connect(), it returns 0 when the connect
is done and thus we don't need the getsockopt() call anyway on Windows.

the patches already applied. No need to apply any patches manually anymore.

ares 1.1.1 is still compatible.

This cures the problem reported by Giuseppe Attardi on October 12, 2003.

(http://bugs.debian.org/212086). Patch by Domenico Andreoli.

multiple DNS lookups can run simultaneously faster. The downside is that
resolving the same host name now can be made at once from multiple threads,
but the upside is that threads now don't alwys have to wait for the others'
resolves. Test case 506 updated accordingly.

an HTTP download, then reset CURLOPT_RESUME_FROM to 0, the next download still
has a Range header with a garbage value." bug report #820502

  addresses, if one of them fails (ipv4-code).

  subjectAltNames when matching certs. This is apparently detailed in RFC2818
  as the right thing to do. I had to add configure checks for inet_pton() and
  our own (strictly speaking, code from BIND written by Paul Vixie) provided
  code for the function for platforms that miss it.

Jeremy Friesner.

sends off the previously set cookie. This also concerns CURLOPT_REFERER.
This fix should stop this.

after a failed transfer.

my_getaddrinfo() has been called (and wait has been set to TRUE), we check
if the name already is resolved and if so don't return wait status to the
parent. This can happen with IP-only names.

properly or risk getting into trouble!

cleanups.

reset the authentication state to make it restart on the next connection.

This of course because NTLM is connection-oriented, whilst all other
authentication schemes are not.

is over or not, as if it isn't we shall not begin any PUT or POST operation.

This cures bug report #805853, and test case 88 verifies it!

NOT free the existing hash entry when we try to add a new one that matches
an existing entry. We now instead free the new one, and make the parent
function use the old entry's struct instead.

"Negotiate" authentication as well.

not used by any code in libcurl!

Problem tracked down by Bjorn Reese.