Thus said Stephan Beal on Wed, 04 Mar 2015 18:35:56 +0100: > For every "round-trip" the credentials of the remote repository are > checked. If no key exchange has been done, the password has to be > entered over and over again.
What version of Fossil are they using? Starting with Fossil version 1.28, sync operations via SSH no longer connect once per round-trip, but instead connect once per sync, as long as the server is also at least version 1.28. Prior to 1.28, if you cloned using SSH and a username/password, you would be required to enter your username/password with each round-trip. This could be annoying with large repositories. Similarly, if you did fossil sync/push/pull for an existing clone, you would potentially have to enter your password multiple times if there were enough changes to warrant more than 1 round-trip. With 1.28 and later, clones now only require entering a password once because the entire clone is done in a single SSH connection. And for sync operations, again, each sync/push/pull requires entering the password only once, even if there are dozens of round-trips required to synchronize the content. > Cloning over SSH > > After creating a local repository, it would be nice to be able to > clone it to a remote location via SSH. Mercurial and Git can do this, > because source and destination may be local filenames or URLs. In > Fossil, the destination has to be local. As a result it is necessary > to copy the repository using scp (which is not always available on > Windows) and later synchronise it with the local one. I can understand this particular use case. It's certainly an good one (if one is accustomed to SSH tools being able to be bidirectional) and basically would mean that the client side would have to switch into ``server mode'' and then the remote side would have to switch into ``client mode'' as far as the clone is concerned. But that also then raises the question... What exactly will the remote-url be on this newly created remote clone? Or is this just intended to be a way to push out a clone from a local machine to a remote machine but without relying on scp? Also, I'm a little fuzzy on how Windows is relevant here. It isn't possible to push a clone to a Windows machine using SSH because, well, Windows doesn't normally have a SSH daemon. And if it does have a SSH daemon, why doesn't it also have SCP? At any rate, this will require some additional thought to come up with a way to make the fossil client actually ``serve'' a clone to the server. Thanks, Andy -- TAI64 timestamp: 4000000054fbe86f _______________________________________________ fossil-dev mailing list fossil-dev@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/fossil-dev
