Serial device locking by convention uses files in a special directory which are named according to the name of the device. Unless all references to the device use the same name, this locking fails. That's why cua devices are no longer used. Why were they ever used? Because old versions of unix had two different devices for the same serial port, one on which an open would hang until the device asserted Carrier Detect (which getty's used) and another which open would return without having Carrier Detect (used for program which wanted to dial out on the modem). In the modern world we have the same device for both uses and control this type of behavior by passing flags to the open system call. For a while the old cua device files were kept around so that old programs which referenced serial lines as cuaX would still work. This however created a lot of problems because the locking convention fell apart unless everyone used the same device name. Thus cua devices were forever banished.
Robert D. Hilliard wrote: > Maybe someone more knowledgable can answer this, I can't. I have > never had any problems with /dev/ttyS? devices in a script. I have > been hearing that /dev/cua? devices are deprecated for years, and they > haven't been included in the debian distribution since rex. > > Bob > > On Tue, 10 Feb 1998, Steve Hsieh <[EMAIL PROTECTED]> wrote: > > > > Use of /dev/cua? may be deprecated, but I can't get scripts to work with > > /dev/ttyS?, only /dev/cua?. Can anone explain why? > > -- > TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to > [EMAIL PROTECTED] . > Trouble? e-mail to [EMAIL PROTECTED] . -- Jens B. Jorgensen [EMAIL PROTECTED] -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
