On 2009-10-06 00:15 , Perry Lee wrote: > How about using one of the time_ variables kept track of by curl? > > For example (from the man page): > > time_connect: The time, in seconds, it took from the start until the TCP > connect to the remote host (or proxy) was completed.
Simultanously opening TCP connections to multiple hosts (and probably even a higher level protocol like HTTP) is not as lightweight as sending a few ICMP echo packets. If there is really someone interested to implement better alternative fetch methods, I would recommend to take a look at external tools already designed for this task before reinventing something new. For example see <http://axel.alioth.debian.org/> In my opinion our current ping based approach is fine enough, as it is now in use since 1.7.0 and we did not get major complaints about fetch failures due to this method until now. Servers which do not respond to the ping request after 3 seconds will be sorted between servers which responded and the fallback URLs. It should always work, only the sort order is influenced by ping. Rainer _______________________________________________ macports-dev mailing list [email protected] http://lists.macosforge.org/mailman/listinfo.cgi/macports-dev
