On Wed, 24 Oct 2007, George Woltman wrote:

> 1) Can anyone explain in better detail the cause of the deadlock and/or
> suggest a workaround?

A work-around would be to build your libcurl to use the "standard" synchronous
name resolver.

> 2) Is there a way to have the resolver not spawn a thread or cache and
> reuse threads?

Not in the current code base, but it could surely be written to do that.

-- 
  Commercial curl and libcurl Technical Support: http://haxx.se/curl.html