gEDA-dev: Re: Bug reports for gattrib

[Steve Meier]

I am going to wade in on this subject. 

The net attribute when attached to symbol allows the user to have hidden
nets. That is a net with no graphical pins. Some people like this
because they can define for example

net=Vcc:14

Which tell the netlister that pin 14 of this device should be connected
to all nets that have a net name of Vcc.

The problem is, that we now have so many different values that Vcc could
be even on one board. Is Vcc 5V or 3.3V or even 1.8V? We really don't
want to unintentionally short all of these possibilities into a single
net.

My solution to support both "the yes short all those nets" and "no let
me define what I want Vcc to be for this device" is to put at least one
graphical pin for Vcc. At netlist time when I find a net attribute I
check to see if any of the pins associated with that attribute are also
a graphical pin.

If there is at least one pin then I override the net attributes netname
and use the net connected to that one graphical pin.


net=Vcc:A15,A16,B15,B16,C15,C16

and my symbol has a pin with attached (to the pin) attributes of
pinnumber=A15 pinlable=Vcc 

Now I have not made use of gattrib but I would expect it to allow me to
edit at the symbol level all of those hidden net attributes and then it
would be cool if it could tell me if that net attribute because it has
or hasn't at least one graphical pin that converts the net from hidden
to associated with that pin.

Thanks,

Steve Meier


On Mon, 2007-04-02 at 12:54 -0700, Stephen Williams wrote:
> Stuart Brorson wrote:
> > On Mon, 2 Apr 2007, Stephen Williams wrote:
> 
> >> The whole point (at least in my case) of this attribute is to
> >> avoid a spray of pointless pins. I think it's much clearer to
> >> think of it as "Oh and by the way, connect GND to pins 5, 7, 9
> >> and 12." I'm not sure what attaching the net to a pin would mean.
> > 
> > The thing is that there is a one-to-one mapping from a pin to a net.
> > (There is not a one-to-one mapping from a net to a pin, OTOH)  That
> > means that the place to display this attribute is under the pins tab.
> 
> That is different from saying that the net= attribute should be
> attached to "pins". "Attaching an attribute to a pin" is not the
> same as "Interpreting an attribute that declares abstract pins."
> The latter is what is happening in my schematic. But the problem
> is that "pins" created by a net= attribute do not necessarily
> exist in the schematic so cannot have attributes attached to them.
>