Steve Youngs <[EMAIL PROTECTED]> writes: > (make-symbolic-link "/[user@host]/path/file" > "/path/symlink") > => FILENAME -> remote; LINKNAME -> local > > (make-symbolic-link "/path/file" > "/[user@host]/path/symlink") > => FILENAME -> local; LINKNAME -> remote > > (make-symbolic-link "/[user@host]/path/file" > "/[user@host]/path/symlink") > => FILENAME -> remote; LINKNAME -> remote
The current implementation produces something strange (an error, perhaps) for the first case, and it creates a symlink pointing from the file /[user@host]/path/symlink to the file /[user@host]/path/file in the other two cases. For the other two cases, this means that logging in as user@host, then doing "ls -l /path/symlink" will show something like "symlink -> /path/file" in the current implementation. I think you want the behavior for the first case to differ: you want the first statement to have the same effect as the third statement. I disagree with both the current implementation and with your suggestion (or what I think your suggestion is). In my opinion, after calling make-symbolic-link with "STRING" as the first argument, "ls -l" should *always* print "foo -> STRING", regardless of what characters STRING contains. So if STRING is "/][][@@@%$@#", then the "ls -l" output should be "foo -> /][][@@@%$@#". Only when _following_ the symlink is it interpreted as a filename, relative to the source of the symlink. This would be aligned with the spirit of the "ls -s" command in Unix, in my opinion. kai -- A large number of young women don't trust men with beards. (BFBS Radio) _______________________________________________ Tramp-devel mailing list [EMAIL PROTECTED] http://mail.freesoftware.fsf.org/mailman/listinfo/tramp-devel
