On Fri, 15 Oct 2004, Karthik Ramakrishnan wrote:

> That is correct - I want to intelligently select from a pool of CURL
> handles. Even if the connection is closed - I get the benefit of the CURL
> handle caching the SSL session ID's. So I can use the curl handle which has
> the cached session ID rather than some other CURL handle which does not have
> the SSL session ID in its cache.

The situation would thus be different once we make it possible to share a
single SSL session ID cache between multiple handles. That's a feature I
personally see as more prioritized.

> Yeah, from what I read EFFECTIVE_URL is just going to return the last URL
> accessed by the CURL handle. Any suggestions on this issue ?

It would of course require a patch of the source code to implement, and I
figure it would need to return a pointer to an allocated struct (containing at
least the number of live connections and then an array with char pointers to
each URL last fetched on each single connection) and the struct would need to
be curl_free()ed after use.

The connection cache as implemented today doesn't keep the "last URL", only
host name, port number and a litte extra data.

-- 
      Daniel Stenberg -- http://curl.haxx.se -- http://daniel.haxx.se
       Dedicated custom curl help for hire: http://haxx.se/curl.html