Article 113623 of comp.os.vms: On 11 Dec 96 23:52:22 GMT, A. Andrew Brennan wrote: > [a] read a local newsgroup, > [b] post a simple test note to the group, > [c] leave the article display, > [d] attempt to re-enter the group and "Selecting group..." > appears, but we never actually see the article display. Is there a time delay between steps b and d? That is, do you wait long enough for the server to timeout or does it occur immediately? >>To avoid the possibility or of slrn_exit_error being called >>recursively, in slrn.c, replace slrn_exit_error by the following >>TWO functions: > > I'll ask the question that has me puzzled ... if it were calling > anything recursively, wouldn't there be some process activity or > some increase in CPU time?? I'm seeing absolutely nothing here. It looks like it might be sitting some sort of wait state while trying to gather information from the server. The slrn unix code allows one to ^G to drop the connection without leaving the newsreader. Is there a way of killing the process from another window to force a traceback? Other than that, all I can suggest is that you put it in the debugger to find out where the problem is. Another thing: the VMS version uses the TCP/IP code that is in the clientlib.c file whereas Unix and OS/2 use the code in sltcp.c. I would have ported that part to VMS but there does not seem to be a consistent VMS TCP/IP socket implementation; that's the reason that the clientlib.c code is such a mess. Incidently, did you make this change to cancel_read in clientlib.c: static void cancel_read (int sig_unused) { #if 1 slrn_error ("CANCEL_READ: NNTP read failed."); #else slrn_exit_error("CANCEL_READ: NNTP read failed."); #endif } Thanks, --John