kicad-developers team mailing list archive

[Vesa Solonen <vesa.solonen@xxxxxxxx>]

30/03/16, 14:24, Paul "LeoNerd" Evans kirjoitti:
> On Wed, 30 Mar 2016 12:22:48 +0100
> "Paul \"LeoNerd\" Evans" <leonerd@xxxxxxxxxxxxxx> wrote:

> Taking a look at e.g. what Eagle does here, they have a 3-way split:
> 
>   *) A single schematic symbol of simply "NPN transistor", with pins
>      named B, C, E.
> 
>   *) A single PCB footprint of simply "TO-92", with pins numbered 1, 2,
>      3.
> 
>   *) A "component" which associates a schematic ("NPN transistor") with
>      a footprint ("TO-92") and provides the pin mapping E=1 B=2 C=3
> 
> This is great, because adding new kinds of NPN transistor means never
> having to draw that schematic symbol again - you just associate it with
> new footprints (or maybe the same footprint with different pin
> mappings). Adding new TO-92-shaped components means never having to
> redraw that footprint - you just associate it with new symbols (maybe
> some 2-pin current sources that come in TO-92 cases with one pin
> unused). Overall, it minimises the amount of work needed to add new
> parts.

This is how it should be, but obviously many (on this list) consider
this to be too complicated. What they fail to see is the flexibility and
automation this kind of clean separation allows. There is a lot of
discussion ad arguing in the list archiver if you have time to browse...

> Who's with me here?

Me, but this particular issue didn't get much anyhere during the years
I've been around. The new schematic and component format tries to solve
the issue by embedding the mapping data with symbols or thereabouts, but
is not as elegant.

Sometimes it also pays to see how others have solved the problem and
learn the best parts of it. I bet library handling has been carefully
considered in the commercial sector, when the libraries play a big role
in customers buying decision.

-Vesa