Re: gEDA-user: DRC vs shorted nets?

["harry eaton" <bumpelo@comcast.net>]

> I've finished a layout, ran the DRC and got this:
>
> 1: Rules are minspace 7.00, minoverlap 4.0 minwidth 8.00, minsilk 8.00
> 2: No DRC problems found.
>
> Seems good.
>
> Then, by mistake, I hit "O" to optimize the rats nets.  There being no
nets
> left, they are all routed, I got this:
>
> 3: WARNING!! net "P3V3" is shorted to net "GND"
> 4: WARNING!! net "GND" is shorted to net "P3V3"

In the future it is good to request a rat's nest at the end of the design
(rather
than rely on it being a mistake) in order to verify the connectivity.
Netlist shorts
are NOT DRC errors. There are connectivity errors which are checked by the
rats nest function.

> I though I would get "Congratulation's all nets are routed"?
>
> Nothing shows a location of the shorted point, if there really is one.

While it's possible that there is a bug indicating a false connection, I'd
say there
is a VERY strong likelihood that there is a short between these nets. There
is no
general solution to answer the question "where is the short?" Which is why
no
answer is given. For power and ground nets, there are so many connections
that
it may be hard to figure out which one(s) don't belong, but you've got to do
it.
If you use the "f" key to see what is connected to a point, then you can
start deleteing
tracks, then hit "f" and see if they're still joined until the connection is
broken.
Use undo to restore to the point before any tracks were deleted, and see if
you can
understand why the final one was responsible.

> I've noticed that I get odd shorts after doing Othropull's but usually
they
> show up in the DRC.

People need to submit example layout to the bug tracker at sourceforge so DJ
can
track down these bugs; several people have reported shorts, opens, even
element
destruction, but no examples have been provided (mea culpa I've seen some of
these
without providing examples either). My feeling right now is that the
optimizer is a dangerous
thing to use, and even the undo for it is not error-free.

Cheers,
harry