Maybe the problem is that pressing Reset on an OSA ICC doesn't do anything
(but that's that same on a TN3270 into VM).

You say that TN3270 doesn't have an ATTN function for non-SNA.  I guess
that's true.  But, PCOMM does allow you to assign a key to [sys attn] and it
can appear on a popup panel as 'attention', and -when connected to VM's
TCPIP Telnet deamon- that key/function brings me directly in CP READ.  In
any case, from a locked fullscreen, from an not locked fullscreen, from
linemode.

I know one can "turn" this off by setting BRKKEY NONE (also part of CONCEAL
ON).  Countless are the VM sysprogs to which I told "don't run like this,
you need the break key in looping conditions" after which I gave a short
demo.

Maybe there's no bug in OSA ICC: in an earlier append I said breaking in was
different depending on the terminal type, and with OSA ICC one has a mix.
It could be fixed if OSA ICC would translate this [sys attn] into an unlock.

2010/3/31 Alan Altmark <[email protected]>

> On Wednesday, 03/31/2010 at 06:47 EDT, Richard Troth <[email protected]>
> wrote:
> > ICC also doesn't support NOHOLD for DISCONN and LOGOFF like some other
> > "real 3270s" and like LDEV.  That's a whole nutha discussion.
>
> ICC has no idea that you have logged off.  Unlike 3270 printer sessions,
> there is no provision in the display data stream for an EOF marker to tell
> the telnet server that the host is finished.  And if there were, we'd have
> to provide a TERMINAL LOGOFF HOLD|NOHOLD command to control it since 50%
> of you would hate it either way.  (I wish I had that already; I always
> LOGOFF HOLD.)
>
> > But ... does PCOMM not have an immediate PA1?  (I'm not used to an
> > ATTN sequence for PA1.)  You wouldn't get the trace, but you'd get
> > control.
> >
> > There's also the trick, I think I learned from Alan Altmark, of
> > putting the cursor back one position from the start of the command
> > line then <Enter> for a VM READ.  It's a hack.  (IMHO)  Your 3270
> > emulator may affect the viability of this behaviour.  You MIGHT then
> > be able to TS.  Alan?  Wanna clarify this?
>
> You're right that TN3270E doesn't behave like a real 3270, particularly in
> the area of Attention processing.  A 3270 control unit can do whatever it
> wants to get the host's attention .  With TN3270, the client is at the
> mercy of (a) the protocol, (b) the telnet server implementation, and (c)
> the interface the telnet server uses to create sessions.
>
> I have to contradict Kris:  TN3270 doesn't have an ATTN function for
> non-SNA emulation (as is done by VM and ICC).   This is why the
> backspace-and-press-enter "hack" was invented; it provides the old-school
> 3215 ATTN function in a "new"-school 3270 environment.  Oh, and the RFCs
> require a telnet server to ignore any telnet IP (Interrupt Process)
> commands it receives on a TN3270 session.  IMO, the 3270 architects did us
> a disservice by not providing ATTN for non-SNA connections.
>
> So while pressing the RESET key unlocks the keyboard, there's nothing the
> TN3270 client can do to get the telnet server's attention.  The ATTN hack
> doesn't work in the fullscreen app since CP isn't doing the 3215-3270
> conversions.  Therefore, if you're in a fullscreen app that has issued
> TERM BRKKEY NONE, then you can easily find yourself trapped (as the author
> intended?) in the app unless you have outside help (e.g. CP MSG/WNG or
> DISC/LOGON).  If you have SET RUN ON, the DISC/LOGON sequence won't help.
>
> If you can get the console into line mode, then you *may* be able to get
> out.  It depends on how creative (annoying) the app was.  But as long as
> you're in fullscreen, you're stuck.
>
> Alan Altmark
> z/VM Development
> IBM Endicott
>



-- 
Kris Buelens,
IBM Belgium, VM customer support

Reply via email to