Greg Klanderman wrote: If some_command uses unbuffered I/O, then yes. The synchronisation will be at the write() level, not at the (e.g.) fwrite() level. The data in output.txt will correspond to the order in which calls to write() on either descriptors 1 or 2 occurred. -- Glynn Clements <glynn(a)sensei.co.uk&gt;

Greg Klanderman wrote: Having looked at the code, and seeing some of the problems with it on NT, I'd definitely agree. Why do you need to redirect the output, rather than just slurping it into an XEmacs buffer and processing it there, either from the process-filter or process-sentinel? Redirection is pretty broken on Windows systems, so you might find this less robust than you hope. Since the cvs messages themselves are always sensible (ie they don't appear intermingled) it would also seem to me that this would remove the worries about synchronisation. Unless someone has seen cvs produce garbled output on a terminal :-) Regards, Malcolm