Okay, I now understand your current concept of an Atomic Part.
The last element is the 3D model.
In my opinion, Kicad should have at the Symbol level, the ability to assign Part Number, Value (or ~(none) ) , Footprint, and 3D Model; all this information contained inside the Symbol.
During routing of the board, the physical constraints come in to play. It becomes obvious that the pin header is going to have to be a right angle pin header. KiCad, as of now, can assign that Footprint a new 3D model; but, now the actual bill of materials is going to need to be changed.
The Schematic is correct, the Footprint is correct, but the 3D part is NOT correct; affecting the BOM generation of the PN/MPN to order.
At this time, KicCad can not force all of the board data and 3D Model data into Eeschema data fields and have have these be the primary control; the primary source of BOM control.
This is the concept of why I wanted PcCvb to remain in place. To many things happen on either side of the design process. I happen to use both methods and think that PcCvb should be brought back as a Stand Alone program to handshake between EeSchema and PcbNew.
I quite liked CvPCB for it’s overview of all shematic symbols and their attributes, but unfortunately it never had a “finished” feel. Copying attributes between symbols has always felt a bit weird to me.
The way the “Edit symbol Fields” spreadsheet works is superiour in some ways to CvPCB because of the grouping of similar components in a tree like structure.
It is however also clearly “not finished”. The way that Libraries could be searched and sorted on things like pin count was very powerfull, but at the same time not very easy to use. (Partly because it was difficult to see from the Icons whether a particular filter was “on” or “off”.
In the “Edit Symbol Fields” there is a directl link in the footprint field to the footprint editor, but when selecting footprints like that, you have to browse trough all the libraries, and then select a part from one of the libraries, and then close the Footprint library browser, which also closes the 3D view of the part, and you have to open it again for the next footprints. This is especially anoying with things like connectors, where the 3D View is a real help in selecting the right part, because just the pads and silkscreen are not much help when searching for the right connector. The names of connectors also do not mean much to me.
Because of the very obvious improvements I just assumed that it’s on some roadmap to get implemented when the developers have time for this.
How do you assign the right angled pin header in this example:
I have never cared much for back-annotation, so instead I just go back to the schematic, assign a new footprint there to the symbol and then press [F8], which updates both the Footprint and the 3D model on the PCB.
I’ve also just been experimenting in Pcbnew with editing a footprint and exchanging it for another. To me it’s very unintuitive to have to press “apply”, and I missed it a few times even though it’s just next to “OK”.
When I change Footrpints this way in Pcbnew, the 3D model is also updated, and keeps in sync with the footprint.
Then I looked through the menu and icons in Pcbnew for a way to back-annotate the changed footprint into the schematic, but could not find it. I do not care for this much either, because I always use the schematic as the master reference for all symbol attributes. I never had a need to use back-annotation.
Because of the Tree like collapsing in the “Edit symbol Fields” spreadsheet it is also easy to change all THT resistors in a design to SMD.
This is either already implemented, or I do not understand you correctly:
Value and footprint have always been assigned in the schematic. “house part numbers” can be easily added as custom attributes, but are impossible for the KiCad team because everybody uses their own house part numbers (This reeks of the integrated worldwide database idea which keeps poppin up).
Only thing left is the 3D model. Now it seems to be linked to the footprint, and that looks OK to me. Why would you want to be able to put a footprint of an SMD resistor to a THT resistor?
And even this can be done, if needed.
What I sometimes do is:
1). Draw some schematic.
2). Start the PCB, and import most of the parts.
3). Discover something is missing.
4). Put a custom (one-off) part in Pcbnew (with custom name, footprint, 3D model).
5). In Eeschema, put the name of the custom part in the “Footprint” field.