[Gnucap-devel] a bug? (was: Re: Unstable inverter isn't unstable)

[Felix Salfelder]

On Wed, Mar 05, 2014 at 10:25:35AM -0500, al davis wrote:
> What happened is that it found a "meta-stable" state.  It's a 
> singleton in the response.

a good opportuniuty to report something i consider a bug...

on the netlist orestes posted i tried

.DC Vg 0 -5 -.5 loop
.DC Vg 0 -5 -5 cont
.DC Vg 0 -5 -5

i get this

#           V(3)      
 0.         0.016234  
-0.5        1.5163    
-1.         3.0163    
-1.5        4.5163    
-2.         6.0163    
-2.5        7.5163    
-3.         9.0164    
-3.5        10.516    
-4.         12.016    
-4.5       -13.013    
-5.        -13.013    
-5.        -13.013    
-4.5       -13.013    
-4.        -13.013    
-3.5       -13.013    
-3.        -13.013    
-2.5       -13.013    
-2.        -13.013    
-1.5       -13.013    
-1.        -13.013    
-0.5       -13.013    
 0.        -13.013    
#           V(3)      
 0.         0.016234  
-5.         0.057836  
#           V(3)      
 0.         0.016234  
-5.         0.057836  

from how i understand "cont", gnucap should have computed

[..]
-0.5       -13.013░░░░
 0.        -13.013░░░░
#           V(3)░░░░░░
-0.        -13.013░░░░
 0.        -13.013░░░░
#           V(3)░░░░░░
 0.         0.016234░░
-5.         0.057836░░

if that is true, i suspect that tr_begin should only be called from DCOP::sweep
in case !_cont. (sorry, i haven't checked for side effects a lot).

cheers
felix