kicad-developers team mailing list archive

[Wayne Stambaugh <stambaughw@xxxxxxxxxxx>]

On 6/4/2014 1:29 PM, Lorenzo Marcantonio wrote:
> On Wed, Jun 04, 2014 at 06:07:18PM +0200, Tomasz Wlostowski wrote:
>> Looking at demos/video.kicad_pcb file, there are modules that use French
>> layer names:
>> (layers Cuivre Dessous.Pate Dessous.Masque)
>> as well as the standard English ones or a mix of both:
>> (layers *.Cu *.Mask Dessus.SilkS)
> 
> AFAIK this is an error, or, better, an old file. Giving the problems
> stated before...  Last time I checked layers ID are the layer names which
> are free for copper ones (the first 16) and fixed in stone for the
> following 16. Number-to-id is specified in the beginning of the file (there
> is a layer form at the beginning with the table)
> 
> So Cuivre is a copper name (usually copper) while Dessous.Pate
> Dessous.Masque are plain wrong, these should be B.Paste and B.Mask (if
> my french is good enough). They are recognized because they are pointing
> to the right 'number' thru the name in the layer form (the real layer id
> is the number, but in the sexp file it's always decoded thru the table).
> In short, it works anyway for old files.
> 
> In fact if you do a 'save as' they change to the 'standard english'
> ones. Ah, by the way, the *. shortcuts are simply hardcoded, there is no
> pattern matching involved.
> 
> If it isn't working that way then something changed and I'm behind with
> the specs. Sorry about that.
> 

There are two separate concepts of layers.  The layers defined in a
board and layers assigned to footprint elements.  When you design a
footprint, you don't necessarily know what board layer(2) the pads are
going to be placed on.  This information gets changed in the module
sexpr when the footprint is placed on the board.  There has to be some
mechanism to ensure that there is not a disconnect between the board
layers and the footprint (module) layers.

There is nothing preventing us from extending the layer sepr definition.
 We could add user defined names to the layer definition without
breaking the existing internal layer notation.  We could also add the
concept of pairing as well.