On 5 November 2010 10:06, Peter Sylvester <[email protected]> wrote: > hi, > > On 11/04/2010 11:35 PM, Hongli Lai wrote: >> >> On Thu, Nov 4, 2010 at 11:19 PM, Peter Sylvester >> <[email protected]> wrote: [...] >>> If one wants to connect to a particular IP address >>> in order to go to https://some.domain/ then the >>> problem could be regarded as a "proxy issue", >>> instead of a proxy that uses CONNECT, one could >>> invent a direct/immediate proxy type. >>> So instead of resolving the DNS for a direct connection, >>> one would use connect to this "proxy". >> >> That would require me to setup a proxy server, and for what gain? Just >> to make an HTTP library do what I want? > > let's take a strawman like > > --proxy ip-address --proxyport 0 or DIRECT or another parm.
I think he just misunderstood what you meant by proxy. i.e. you meant a sort of virtual proxy that only lives inside libcurl's head, but he thought you mean e.g. installing Squid. > or > --resolve host.example.com:ipaddress I prefer this since to me it seems more like what is actually required. You want to connect to host.example.com, but you want to do the DNS resolution manually instead of through a normal resolver. -- Michael Wood <[email protected]> ------------------------------------------------------------------- List admin: http://cool.haxx.se/list/listinfo/curl-library Etiquette: http://curl.haxx.se/mail/etiquette.html
