Re: gEDA-user: Strange behavior with input-1.sym and output-1.sym?

[John Doty <jpd@xxxxxxxxx>]

On May 7, 2009, at 12:19 PM, dfro@xxxxxxxxx wrote:

>
>
> John Doty wrote:
>> On May 7, 2009, at 6:53 AM, Stefan Salewski wrote:
>>
>>> Of course it may be a bug of 1.4.0, I
>>> can not test this, have only 1.4.3 available.
>>
>> One could make the case that the bug is in 1.4.3. A component without
>> either refdes= or  graphical=1 might reasonably be considered an
>> error. But the treatment of attributes in gEDA keeps changing. I wish
>> there was better documentation on how attributes are actually *used*
>> by gschem and gnetlist. The Symbol Creation Guide is "style manual",
>> not  a language definition.
>>
>> The "anything goes" approach to attributes was fine when there were
>> few special cases in the core code, but increasingly I'm seeing
>> strange behavior that's apparently due to the core code assuming
>> attribute meaning that's undocumented and/or specific to the gsch2pcb
>> flow. This is not good.
>>
>> John Doty              Noqsi Aerospace, Ltd.
>> http://www.noqsi.com/
>> jpd@xxxxxxxxx
>>
>
> And, if it becomes part of the gschem/gsch2pcb/pcb standard

There is no such thing. There is a standard for the files used by  
gschem/gnetlist, and a separate standard for pcb (which has a longer  
history). It is gsch2pcb's job to translate (using gnetlist). But the  
same schematics should (as far as practical) be translatable to other  
kinds of netlists, to be handled by other back end tools. Features  
specific to the gschem/gsch2pcb/pcb flow do not belong in gschem or  
the common part of gnetlist.

> that all
> non-physical schematic objects require 'graphical=1', then  
> 'graphical=1'
> should be automatically part of these objects (i.e. input-1.sym,
> output-1.sym), so that the user does not have to add it every time
> he/she uses the symbol.

Indeed. The problem here is that graphical= is a magic attribute to  
gnetlist, but what it does is not completely defined. Historically,  
it seems to have meant "don't include this component in the netlist".  
So, something like a title frame has graphical=1. But connections to  
pins on a graphical component still do something, through the net=  
attribute. That attribute is confusingly overloaded: it's used to set  
the name of a visibly attached net, but also to attach an invisible  
pin to an externally named net. And it depends on another confusingly  
overloaded attribute, pinseq.

>
> Also, I think this info about the attribute treatment of
> physical/non-physical schematic elements should be added to the  
> gsch2pcb
> tutorial.

It's more a gschem issue, since gsch2pcb is only part of one possible  
flow, but the issue really applies to all flows.

 From my perspective, your use of these symbols to name nets seems  
strange. I think of these as hierarchical connection devices. To name  
a net it is simpler and less confusing to use the netname= attribute  
rather than a symbol, I think. But it is again unclear what these  
symbols were intended for by their original authors.

>
> Dave
>
>
>
> _______________________________________________
> geda-user mailing list
> geda-user@xxxxxxxxxxxxxx
> http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
>

John Doty              Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd@xxxxxxxxx




_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user