This is for a Ruby library I'm writing, hence the reason for passing the long value instead of using the constant. The library is here: http://github.com/pauldix/typhoeus
The relevant c code is here: http://github.com/pauldix/typhoeus/blob/cc977ca8249825a371b76b1d6eb1e932a7cbe1b3/ext/typhoeus/typhoeus_easy.c And the Ruby code that interfaces with that is here: http://github.com/pauldix/typhoeus/blob/cc977ca8249825a371b76b1d6eb1e932a7cbe1b3/lib/typhoeus/easy.rb And the example where I call the ruby code is here (lines 17-23): http://github.com/pauldix/typhoeus/blob/cc977ca8249825a371b76b1d6eb1e932a7cbe1b3/spec/typhoeus/easy_spec.rb I'm not sure if that helps at all for narrowing it down. I'd appreciate any ideas. Thanks, Paul On Mon, May 4, 2009 at 4:28 AM, Daniel Stenberg <[email protected]> wrote: > On Fri, 1 May 2009, Paul Dix wrote: > >> curl_easy_setopt(easy_handle, 155, 500); // for say 500 ms timeout. >> where 155 is the CURLOPT_TIMEOUT_MS value > > Why on earth would you use 155 and not CURLOPT_TIMEOUT_MS ? > >> What I'm finding is that it automatically rounds up to the nearest second. >> So it's really only doing full second timeouts. So a ms value of 1001 would >> actually do timeouts at 2 seconds. I saw that I need to have it built with >> c-ares and based on the version returned by curl_version, I think I do. > > If that's really the case it seems like a bug somewhere. Can you post us a > full example that we can try with? > > -- > > / daniel.haxx.se >
