On Dec 1, 2005, at 7:10 PM, Dan McMahill wrote:
Not that I know of, but that and also cross probing (click a gschem
net to plot it, click a gschem element to select the corresponding one
in pcb) would be super useful.
It seems like you'd want to put an annotation attribute on each pin to
let a simulator write a voltage to those. And on elements like a MOS
transistor, have an array of annotation attributes (or maybe just a
single multi-line one?) where you could write back things like gm,
cgs, etc.
The reason I mention cross probing is that that and the backannotation
are both things where you'd like some other program to communicate
with a running gschem and have it do something.
Anyone have experience with such communication between processes?
Hopefully if it ever gets implemented we don't open up a security hole
by listening on some socket. Not that any commercial tools (*cough*
cadence) have ever had a DOS vulnerability due to this...
I've done this sort of interprocess communication, both with and
without sockets. For a socket-free implementation I would suggest
using a combination of shared memory and semaphores. Define a protocol
based on passing structs back and forth using the shared memory
segment, and use semaphores for the processes to synchronize and notify
each other.
-Dave
--
Dave McGuire "You'll have to be a lot more specific than
'that
Cape Coral, FL girl last night.'" -Ted McFadden
