http proxy is complicated

There were a post that gdata-python-client had no proxy support on GData Python Client Library Contributors.
This week I had implemented the proxy support of that library and posted a patch. Jeff had looked into it and got it into the svn repository.

Not until I wrote this code I was not aware of the complexity of http(s) proxy mechanism. If the final destination is http url, things are very simple. To acomplish the duty, just connect the proxy server and issue the http method as usual but with absolutely URL(started with 'http://') instead of partial URL(started with '/'). But the destination becomes https url, things get complicated. We have to use CONNECT method to open a ssl socket to dest server directly. And then issue the http method on this ssl socket.

Ah... explanation in English is very difficult.

Anyway, I had learned little about proxy mechanism :)