carl@CITHEX.CALTECH.EDU (03/15/87)
The documentation for SET TERMINAL/SECURE_SERVER reads:
/SECURE_SERVER
/NOSECURE_SERVER(default)
Determines whether the BREAK key on the terminal logs out the current
process. With /SECURE_SERVER in effect, pressing the BREAK key when there
is no current process initiates the login sequence. By default, the BREAK
key has no effect.
The /SECURE_SERVER qualifier has no effect on terminals that are set with
/AUTOBAUD.
If this were correct, then there would be no point to using the /SECURESERVER
characteristic in situations where there is uncertainty as to whether the
terminal is capable of sending a break: if /AUTOBAUD is set, it does nothing,
and if /NOAUTOBAUD is set, the /SECURE_SERVER characteristic REQUIRES a break to
initiate a login sequence. Where might uncertainty as to ability to send a
break originate? Well, in my case, it's due to using Ungerman-Bass NIU's in a
local area network. In order to send a break from a terminal connected to my
VAX via NIU, both the NIU at the terminal end and the NIU at the VAX end have to
be programmed to transmit breaks. Thus, even it the VAX-end NIU is so
programmed, I can't count on the terminal being able to send the break.
However, all is not lost! The documentation is WRONG. The effects of
/SECURE_SERVER and /AUTOBAUD are as follows:
Terminal Setting Effect of BREAK
--------------------------- --------------------------------------
/NOSECURE_SERVER/NOAUTOBAUD BREAK is ignored
/NOSECURE_SERVER/AUTOBAUD BREAK is ignored
/SECURE_SERVER/NOAUTOBAUD BREAK disconnects current process, and
is required to start a new login
/SECURE_SERVER/AUTOBAUD BREAK disconnects current process, but
is NOT required to start a new login
The documentation, by the way, also fails to take into account virtual
terminals, in that break will under no circumstances log out a job on a virtual
terminal; it does the equivalent of the user typing a control-y then the
"DISCONNECT/CONTINUE" command EXCEPT that it works even if the user has issued a
SET NOCONTROL=Y command! All in all, the /SECURE_SERVER terminal characteristic
is much more reasonable and useful than the documentation suggests. My system
is, by the way, running VMS 4.5, and I can make no guarantees about how these
characteristics interact under other versions of VMS, and would appreciate
hearing from anybody who actually KNOWS whether the documentation or the
implementation is to be regarded as incorrect (i.e., will I have to set all my
terminals /NOSECURE_SERVER when the next release of VMS comes out?).