Hrvoje Niksic <hniksic(a)iskon.hr&gt; writes: No, the problem is evidently what was just observed by Michael, about the tab instantiator error; I just hadn't gotten as far as peeking into the returned buffer. In any event, "gnuclient -nw" should not (and, now looking at the code more carefully, would not) exit immediately, because in the -nw case, gnuclient does not ship across the 'quick qualifier on the gnuserv-edit-files directive. The really perverse situation is that, in the -nw case of immediate departure, XEmacs tries to retain control of the tty anyhow. I was in fact exploiting this with "gnuclient ; sleep 1000000" so that XEmacs didn't conflict with my shell. I suspect that a tweak to XEmacs is in order, to force giving up control of the client tty in the case of any gnuserv-edit-files error return.