Hello, While trying some of the new IPv6 related changes, it dawned on me that Fossil clone/sync using SSH has had IPv6 support inherent in the SSH client all along. So I tried it and it worked flawlessly, with one exception. The information it printed out was wrong as it told me that the IP address to which it connected was an IPv4 address, when in reality it was IPv6 (I could easily tell because my SSH client told me that I had connected to the host via an IPv6 address).
I think this behavior is wrong (and am disappointed that I introduced it in the first place). I think a better behavior would be just to simply log the URL name let the user figure out what IP it connected to if they care: http://www.fossil-scm.org/index.html/info/52f08008e2790a8140d947a680bb1c73d9bc1f5d Another alternative would be to do something similar to the standard socket connect and lookup the hostname and try each returned address in sequence (thus duplicating what SSH already does in Fossil). This would allow Fossil to reliably know what IP address it connected to, but it would mean calling SSH with a different IP each time through the loop, for example: ssh user@192.168.5.6 fossil http ssh user@fe80::22f:a0ff:fea7:1afb fossil http And so forth, however, this would not be without side-effects either. If the user has custom entries in their ~/.ssh/config for the hostnames that they use with Fossil, the attempt to be smart about figuring out the IP would invalidate their configuration (e.g. a Host section that overrides what the username is when connecting). Any thoughts? Is the commit above sufficient for Fossil's purposes? Or shall I go down the path of making Fossil be more diligent in obtaining the IP address? Or something entirely different? Thanks, Andy -- TAI64 timestamp: 4000000054c74372 _______________________________________________ fossil-dev mailing list fossil-dev@lists.fossil-scm.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/fossil-dev
