Cornering the Beaste

[Tom Droege <tdroege@veriomail.com>]

Arne has a problem with his Mark IV in that the first time through his 
program it fails to see the block done.

I have traced this to the register bit that selects the high PROM program 
being set when it should not be set.  I have now boiled this down to a two 
instruction loop that fails every time in exactly the same way that Arne 
sees it fail.  i.e. if fails the first time through and then works properly.

OK, I declare that this is software.  I may try a little more, but Chris is 
in as good a position as I to look for this bug.  Probably we do not even 
want to look for it.  Better to write new software that is bug 
free.  ;^)  Well, new software for which there are experts dedicated to 
maintaining the code.  This is not me.

I admit that I was worried that it was hardware since the bit that sets the 
PROM selection register is next to the Open0 strobe line.  But I think I 
have pretty well ruled out any kind of cross talk.  Why should it cross 
talk only the first time through the program?

I think I will go back to working to make the hardware better.  There is 
lots of room for improvement there.

Arne, the reason I have spent time on this rather than the Dec. clamp 
problem is that I do not understand this bug, but do understand the Dec. 
clamp.  Ok the beaste seems to be in a corner.  We know it is there and any 
mysteries can be blamed on it.  Thus I will go back to working out a 
sequence that will prevent the Dec. from moving before the clamp is on.

Tom Droege