gEDA-bug: gnetlist

[geda-bugs@seul.org]

Full_Name: Mark Whitis
Version: 20040111
OS: linux
Submission from: (NULL) (65.40.216.240)


gnetlist generates three error messages that do not provide any clue as to how
to
find the source of the problem.
 - "Missing Attributes (refdes and pin number)"  (s_net.c)
 - "Could not find refdes on component and could not find any special
attributes!"
    (s_traverse.c)
 - "Found a cpinlist head with a netname! [GND]
These error messages are in desparate need of a file linenumber!

The third error appears to be normal, as it is mentioned in the gsch2pcb
tutorial, but in this case was an error in the board file, since I don't have
any net named
GND.   It was some leftover GND:xx nets in an old symbol create from a
template.

In the first case, I eventually found that I had a "dgnd" symbol that was
missing a pinnumber.  For the benifit of anyone else encountering the same
message, this was found by looking at what was connected to "U?" in the netlist
and looking on the
schematic to see what connected to what U? connected to.

Now, however I get large numbers of the second error.   Looking at the netlist
for U? shows that there is one connection to U?-1 in VCC and one in U?-1 in
DGND,
which gives no way to track these errors back to their origin, though they look
suspicously like there is no problem in the source file at all and it is
improper
handling in gnetlist.

I am not, and can not, use the power symbols supplied with gschem and due to
another gschem bug (update component), if i tried to temporarily substitute
those symbols I would get a royal mess of duplicated refdeses.

The only significant difference between my power symbols and the gschem ones is
that mine have a pinlabel attribute which only one of the more obscure gschem
power symbols (gnd-2.sym) has.     Making a copy of the file with all the
pinlabel attributes stripped of the power symbols did not seem to help.

There are exacly two references to U? in the netlist, one connecting U?-1 to
DGND and the other connecting U?-1 to VCC.  There are only 6 one pin components
(test points)
and those are all accounted for in the netlist.

here is my dgnd symbol, (typed in by hand since I can't copy and paste between
machines):

v20040111 1
P 200 200 200 150 1 0 0
{
T 70 140 5 5 0 1 0 0 1
pinnumber=1
T 260 130 5 5 0  0 0 1
pinlabel=DGND
T 120 140 5 5 0 1 0 0 1
pinseq=1
}
T 550 120 8 10 0 0 0 0 1
net=DGND:1
L 200 110 160 150 3 0 0 0 -1 -1
L 200 110 240 150 3 0 0 0 -1 -1
L 200 110 240 150 3 0 0 0 -1 -1


The problem can be reproduced by drawing a symbol schematic with gschem's own
symbols: 
   - R1   resistor-1 connected between vcc-1 and gnd-1
   - R2   resistor-1 connected between vcc-1 and gnd-1