* Robert Weiner <[email protected]> [2020-10-09 07:31]: > Sounds llike it might use the directory if the buffer from which you > activate the button rather than the dir of the button.
So it uses the current directory of Emacs. Isn't that pointless? If there is current directory, Hyperbole executes exec-shell-cmd in shell by first cd-ing to the anyway set current directory, whatever it is. And I tested it with tramp, there is no consistent behavior: - if there is *shell* buffer, it would not create new shell in Tramp on remote machine, and IT DOES NOT do any "cd" - if there is no *shell* buffer, it creates new shell in Tramp and tries to execute it on remote machin In my opinion doing "cd" in the existing shell for Hyperbole is disturbing whatever user was doing in the shell, user would need to go back to previous directory after Hyperbole executes, I hope you understand. In that case, maybe is better using pushd/popd, but then again, somebody can change shell to something else what is not bash, maybe best would be to invoke new shell buffer for Hyperbole only. Maybe it should be how you said it, switch to directory of the buffer, not to current directory. If I have global button to unmount the disk and eject it, I am not sure if it is supposed to work on remote machines and devices which I would be using by chance, as I am often connected to mobile devices and other computers in the ethernet network. Or maybe it should I don't know, I just did not make commands for other computers. Watching DVD global button is for my computer, not for remote computer. Then am I supposed to first switch, cd, to some local directory before invoking global button, it should not be so. I hope you understand the conflicts. Other question is, if Hyperbole would invoke remote command if it finds remote Hyperbole directory buttons, that would be good.
