That is true, this is in fact very natural behavior, and after further checking I found that this is in fact also the actual behavior.

My problem was that I was using command line parameters to connect and synchronize on startup, but I was not using the name of the stored session as a parameter, but the actual user@server.domain. That way, when I started the program without parameters (selecting the session manually), the correct directories were loaded by default, but using the shortcut resulted in the root directory being selected on the server (in the sync dialog).

Changing the session parameter to the name of the stored session fixed this problem.

Thanks for the pointer, I would not have figured this out without your advise.

Current working directories are used for synchronization. I believe that it is quite natural, isn't it? :-)

The additions to the synchronization in 3.7.6 are excellent. Thanks, I really find the automatic synchronization useful. Storage of settings has also been greatly improved.

However, it seems that the remote directory, which can be set for each connection, is not used when synchronizing, and that neither can a special synchronization directory be saved, even when one selects "use these settings next time"

The problem is that I have a subdirectory that I sync (not my entire remote file area), and sometimes I forget to change the remote directory, leading to the program deciding to copy a lot of stuff to the wrong place, leading to tedious cleanup operations.

Thanks again for a great product!