connection to avoid re-use of it, since we cannot know in what state it is.

existing error is the one returned back

properly bail out from, when using ares.

I guess it seldomly happens on linux and that's why it wasn't found before.
He used Solaris to notice it.

I took the opportunity to rewrite the Curl_connecthost() slightly to feature
less duplicate code in the two different versions (ipv4/ipv6).

This seems more appropriate, and I believe mozilla and the likes do this.

.html files now get text/html as Content-Type.

Pointed out in bug report #839806.

which works nicely since the build dir is before the source dir in the
include path.

in the 229-reply servers respond on a EPSV command and bail out better if
the reply string is not valid RFC2428-compliant.

belong in the public header file. They are not in our name space so we
should not set them globally.

list, we get the time in GMT and not localtime

isn't. In fact, it might be, but we have no idea.

us some trouble.

syntax: file:/path. I added test case 203 to verify this.

outside its given buffer. Discovered and reported by James Bursa.

a range of a document beyond its size, caused libcurl to "hang" until the
server closed the connection and then it returned error 18.

This is bad. This way, we don't return any error at all, which isn't nice
either, as we need to alert the app somehow that the request range was out
of size.

and gcc, to set the -mimpure-text link flag for linking the lib better.

run out of memory.

done even if the result is ignored, as some sites seem to require that.

we might not get a body when we get a 401 with a set of WWW-Authenticate:
headers. This fixes the problem Kevin Roth detected in 7.10.8-pre4 and pre5.
Verified by test case 91.