(Wow, this thread is getting long! Sorry about that)

Jon Perryman is still endeavoring to help me, which is appreciated:

>The DEFINE GRAF did nothing useful. SECUID is associated with the VM
>user's CONSOLE definition. If you look at the z/OS syslog, you will
>find the V CN(*),ACT did not come from FFF. It will show the command
>came from another console that is either defined in the VM CONSOLE
>statement or SYSCONS / HWCI. I'm guessing from the VM CONSOLE defined.

Right, I never thought the VARY came from FFF. I thought (and still believe) 
the VARY made FFF "active" in some sense, since without it, no SECUSER info; 
with it (after the VARY), SECUSER info.

>If you IPL z/OS without the use of a CMS exec, then you will need to
>specify the SECUID on the VM user's CONSOLE statement. If you use an
>exec, then issue the SET SECUID before the CP IPL command.

Of course. I'm doing that.

>Keep in mind that VM CP is about hardware emulation (not about an OS).
>Even a CMS user requires CONSOLE 009 for the IPL CMS to be successful.
>Always think how hardware behaves regardless of the IPL'd os.

As I've said, I've been using VM for over 40 years, so I know this; what I 
don't understand is how z/OS deals with consoles, since it has its own 
full-screen support.

>Realize that you issued the V CN(*),ACTIVATE from a z/OS console that
>was not active. This is not possible in the real world. You would have
>issued the command from an active console. The VM CONSOLE is leaving
>the z/OS console in a funky state or real consoles are also in this
>funky state when powered on with commands enabled but messages
>disabled.

OK, what does "active" mean in a z/OS context? Remember, what I'm trying to do 
is to "connect" (or perhaps "activate") a z/OS console at IPL such that system 
log output is copied there in line mode so the SECUSER sees it. So this seems 
like a key bit of understanding. I keep reading z/OS doc but it's full of 
assumptions and short on examples.

>When you autolog the z/OS user, the VM CONSOLE is in a disconnected
>state. z/OS will not send messages to any console that is powered off.
>In the past, issuing commands was also disabled but that seems to have
>changed.

"powered off"? As in, not active?

>As I said before, you should be able to test this by autologing a CMS
>user that runs an exec that displays messages. It should do something
>similar as z/OS when disconnected. In the case of CMS, it should go
>from RUNNING to some sort of wait.

Yes, of course. I know how that works, have written and supported many products 
that use SECUSER. But they weren't z/OS, weren't running in CONMODE 3270.

>Adding a SECUID (secondary user) sends all messages (CP and underlying
>OS through VM user CONSOLE) to the secondary user. What is the
>hardware state if the secondary user is not logged on or disconnected?
>Does CP simply discard messages or does the hardware act as if it's
>disconnected?

What is the hardware state of...what? If the userid is not logged on at all 
(disconnected *is* logged on as far as VM is concerned, just not connected), 
the virtual console is still there, is still used. But I don't think that's 
what you meant.

>If the disconnect is forwarded from the secondary user, then you will
>need to start the SECUID running PROP prior to starting the z/OS user.

Again, not sure what this means--does "disconnect is forwarded" mean "a CP 
DISCONNECT command is issued by the SECUSER"?

>You say z/OS messages are wrapping. "these" refers to additional
>information about the message. Unlike VM, z/OS can optionally display
>more than the message (e.g. jobname, jobnum and more). z/OS console
>ignores the hardware definition and queries the device to format
>messages according to the device. Since a 3215 is 132 bytes wide,
>expect lines to wrap on an 80 byte terminal if the line exceeds 80
>bytes. I don't know your requirements but lets say only displayed the
>message without anything else, then primary messages would not wrap
>because they are 80 bytes or less. Secondaries would still wrap (e.g.
>D NET command output). Get help from a z/OS sysprog who is familiar
>with z/OS syslog / consoles / messages and discuss what you
>requirements. Since you probably use TN3270, why not define your
>screen with lines longer than 132 bytes.

Well, I'm not worried about ME seeing the lines, I meant that it seemed like 
they weren't going to come to the SECUSER as separate lines. But as I said 
(somewhere), I might should tinker with that more.

This:
>Get help from a z/OS sysprog
...is of course the real solution here. I don't have one handy, alas. Though I 
have asked our service provider for their thoughts, no response yet (I just 
asked, had thought this would be easier than it's turned out to be).

>> I need to define some virtual device at the right address

>Your VM CONSOLE address is correct because you saw z/OS messages that
>can only come through the VM CONSOLE address. Regardless of the OS,
>you will not see anything other than CP and this device from the VM
>user console. For instance, DEF GRAF allows you to dial into the GRAF
>address but has nothing to do with the VM user console.

>The underlying OS uses VM user console as defined in the VM CONSOLE
>definition.

>>CMS users don't run with TERMINAL CONMODE 3270.

>It's true the VM CONSOLE definition for CMS specifies 3215 but CMS
>queries the device. It ignores the hardware definition when required
>otherwise things like xedit wouldn't be full screen.

True. But VM full-screen mode is not CONMODE 3270. Those are different things. 
VM does full-screen via DIAG 58, with CP doing the hard part.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to