I would suggest that a layer is really more of a conta= iner than anything else, it is a way of collecting and grouping objects.

I would not c= all that a layer at all. A layer represents a hunk of copper, epoxy glass, = ink, etc. What you=E2=80=99re talking about should be called a =E2=80=9Cpat= tern=E2=80=9D, I think: it edits the arrangement of material on one or more= layers.

The t= erm "layer" is a well established term for referring to groups of= =20 objects with a common "z-order". A pattern, as you describe it, i= s also a collection of=20 objects with out the restriction of those objects having a common z-order. = Both are collections of objects. I realize I neglected to include "com= mon z-order" in my original statement, apologies. Thank you for helpin= g me to be more precise with my language.

Layers do not always have = physical analogs. When I'm designing boards, using those fancy expensiv= e commercial=20 packages, I will always include what I call an assembly layer that=20 contains instructions for the technician that assembles the board. That lay= er doesn't end up being physically realized at all. A layer only repres= ents something with physical significance after it has been assigned that s= ignificance.

I suggest that at the base level, the software be igno= rant of that significance. This allows a layer to signify whatever you want= it to. Then the layers are given significance when interpreted, by DRC, a = fab, or whatever. If you turn off all DRC you should be able to draw anyth= ing on any layer, regardless of whether or not it can be physically realize= d. What will pass DRC is a function of the fabrication technology and is su= bject to change as technology changes.

Ultimately the group of objects is interpreted in a particular way when a board is fabricated, perhaps as copper, perhaps as a board outline,=20 perhaps as a silkscreen... or perhaps as something else entirely.=20 Anyway, the point is that the concept of a layer is really a more=20 general abstraction and the physicality of them is an interpretation.

<= /div>

Nicklas's statement is true: "Every object drawn end up = on a layer as one or more drawing primitives.", but he is referring to= an object as an individual drawing primitive, not a collection of drawing = primitives. A "pattern" may end up placing many primitive objects= on many different layers, but all of those primitive objects end up on lay= ers.

As to whether or not it makes sense to associate a pattern wit= h a layer, that is an interesting question, and I believe I could make a go= od argument for an implementation of patterns in which such association wo= uld make sense. If your pattern is defined not with absolute layer definiti= ons, but with relative layer definitions, then assigning the base layer of = a pattern to a particular board layer could make sense. Each layer of the p= attern is assigned to a board layer according to the assignment of the base= layer. As an example, consider a pattern with 3 layers, -1, 0, and 1. Assi= gning such a patter to layer 5 of a board would map the three pattern layer= s to layers 4, 5, and 6 of the board, or layers 6, 5, and 4 if the pattern = is mirrored. Such an implementation could offer a great deal of flexibility= . Yes, it is not a complete theory and so has problems, but I think they ar= e problems that could be solved elegantly by refining the idea of how a pat= tern specifies the layers on which its primitives (or sub patterns) are ass= igned.

--Chad