Mail Archives: geda-user/2015/12/09/02:29:00
On Tue, 8 Dec 2015, John Doty wrote:
>
> On Dec 8, 2015, at 5:29 PM, gedau AT igor2 DOT repo DOT hu wrote:
>
>>
>>
>> On Tue, 8 Dec 2015, DJ Delorie wrote:
>>
>>>
>>>>> So already the refdes isn't a unique identifier for a symbol...
>>>>
>>>> In such case it is a slot and the slot number could also be used?
>>>
>>> Sometimes. Other times, like large MCUs, there are no slots, just
>>> different symbols using the same refdes, which the netlisters just
>>> merge together.
>>>
>>> At best, "refdes + pinnumber" may be a unique identifier, *except*
>>> that changing the refdes and/or pin numbers changes the identifier...
>>> and those are the things we want to change.
>>
>> Refdes+pin sounds like a good identifier on PCB-side, but not on gschem-side.
>>
>> When I use stock dual-opamp-1.sym (slotted) pin+refdes it's not unique in the sense that there are two instances of the very same power pins (4 and 8) on the schematics.
>>
>> If I connect Vcc to pin 4 on slot 1 and gnd to pin 4 on slot 2, I get the following netlist (in the gsch2pcb flow):
>>
>> Vcc U1-4
>> GND U1-4
>>
>> Same happens without slotting. However, if I take only one slot and connect both Vcc and GND to the same pin 4, only one net remains:
>>
>> GND U1-4
>>
>> This suggests the pcb gnetlist backend (or maybe gnetlist) already uses a per gschem-component identifier and refdes or refdes-pin doesn't fully identify a component or a pin.
>
> What happens in this case depends on how the back end retrieves the connection information from the core. My check-duplicates back end will detect such cases. http://www.gedasymbols.org/user/john_doty/tools/check-duplicates.html
>
Clear, and I do understand and respect your point of view on why this is a
good thing. However, my opinion on what the desired behavior should be
differs, and I find the current situation (not this specific example, but
the whole "we don't want to have a common way of identifying things") a
bad thing that does increase confusion big times for a marginal gain on
flexibility.
- Raw text -