Mail Archives: geda-user/2013/07/10/18:33:29
On 07/10/2013 03:13 PM, John Doty wrote:
> On Jul 10, 2013, at 3:11 PM, Dave Curtis wrote:
>
>> Working on a symbol generator.... gsymcheck gives warnings on pins that do not have a pinname attr. Is there any real issue with anonymous pins that can cause problems? Or can I just leave off the attr? Or....should I do something like attach pinname=anon3 as a hidden attr when I don't want to display a pin name?
> I believe you mean "pinlabel".
yes... my bad
> The only place it's required is when the symbol has one or more "source" attributes. In that case, it associates the pin with the appropriate IO symbol in the source schematic.
ah ha... OK good.
>
> In general, there are *no* required attributes in gschem. See the IEC 60417 library for a collection of attribute-free symbols. What you need depends on what you're doing with the symbol. Gnetlist needs "pinnumber" on pins, and usually needs "refdes" on symbols. Slotting requires "numslots", "slot", and "pinseq". Some gnetlist back ends also need "pinseq". DRC wants "pintype". "footprint" and "value" are often useful. "device" usage in the symbol library is inconsistent, but the "spice-sdb" back end uses it to figure out how to translate "refdes" for SPICE.
I've covered all of those. All that slotting stuff is one of the big
motivators for 'correct by construction' symbols. I'm not expecting
'footprint', though, because I usually glue that on later with gattrib.
('expecting' as in warning the user that it is missing in the symbol
specification...)
Anyway, thanks for the clarification on pinlabel, my anonymous pin
strategy seems good to go.
I just passed a major milestone with my tool -- I tried an new input
file and I didn't have to fix any bugs to get the output that I wanted
:) so maybe in a few days I can get a git push done.
-dave
>
> John Doty Noqsi Aerospace, Ltd.
> http://www.noqsi.com/
> jpd AT noqsi DOT com
>
>
>
- Raw text -