[Simh] hp2100 bug in BACI/telnet

[J. David Bryan]

On Tuesday, September 30, 2008 at 0:29, I wrote:

> I think that's a bug in QCTerm; the CTRL+O is being stored as part of
> the screen buffer when you type it.  A real 2645 terminal stored typed
> control characters in the screen buffer only when "display functions"
> was on.

I've investigated this a bit further by hooking two (real) HP 2645 
terminals together.  My second statement above isn't quite correct.  Some 
control characters are stored in the screen buffer, while others are not.  
In particular, CTRL+N and CTRL+O are stored, as they switch in and out of 
the alternate display character set.

However, CTRL+O is stored only if it has a corresponding CTRL+N preceding 
it in the line.  A "bare" CTRL+O is not stored.  So:

  A line with <CTRL+N> alternate <CTRL+O> characters

...will be stored and read back from the screen as shown.  However:

  A line with <CTRL+N> alternate <CTRL+O> characters <CTRL+O> again

...will be read back without the second CTRL+O.  This is the bug in QCTerm, 
as both CTRL+Os will appear in the screen data.  And that's why CTRL+O to 
duplicate a line does not insert the CTRL+O in the screen data: because it 
has no corresponding CTRL+N preceding it in the line, the HP terminal does 
not store it, and so does not read it back in screen mode.

A bug report to AICS Research would be in order.

                                      -- Dave