set to "".  This frees the application from having to know which encodings
 the library supports.

the new socket to non-blocking

DATA

 EAGAIN but only EWOULDBLOCK, which caused badness on HPUX. We also check for
 and act the same on EINTR errors as well now.

extract the correct set of cookies to send. This functionality is verified
by test case 62.

portability. Also use Andy Cedilnik's compiler warning fixes.

as otherwise we'll go nuts

  o Save domains in jars like Mozilla does. It means all domains set in
    Set-Cookie: headers are dot-prefixed.
  o Save and use the 'tailmatch' field in the Mozilla/Netscape cookie jars (the
    second column).
  o Reject cookies using illegal domains in the Set-Cookie: line. Concerns
    both domains with too few dots or domains that are outside the currently
    operating server host's domain.
  o Set the path part by default to the one used in the request, if none was
    set in the Set-Cookie line.

RFC1738, which means it'll use one CWD for each pathpart.

 when using the multi interface (too).

 return CURLE_OK no matter what happens.

also now attempt to detect and return the specific CACERT error code.

I edited somewhat and removed trailing whitespaces.

chunked encoding.

 there seem to be non-compliant FTP servers out there that return 200,
 so we accept any '2xy' response now.

Ralph Mitchell's notification.

slash after the host name, but still a ? and following "parameters".

made the Last-Modified (faked) header look correct using GMT always

miss the first letter

and 12 following hexadecimal letters, which seems to be what IE uses.
This makes curl work smoother with more stupidly written server apps.

Worked this out together with Martijn Broenland.