kicad-developers team mailing list archive

[Martin Schreiber <mse00000@xxxxxxxxx>]

On Sunday 15 January 2017 12:10:20 jp charras wrote:
>
> My question was more why do you want to use an helper internal method
> (which needs the initialization of many parameters, made by a top level
> method), rather than the method which generates the drill files as used in
> the demo gen_gerber_and_drill_files_board.py.
>
I learned that to use lowlevel access it is more work maybe but it often 
allows to made better products. :-)
For the documentation stack generator MSEkicadBOM allows to set filename, 
scale, orientation, mirroring... of the individual items. There I use 
PLOT_CONTROLLER and rename/move/delete the produced files.
For the drill files I saw CreateDrillFile() and thought that it is a handy 
function to generate individual files directly.
Maybe it should be documented which pcbnew.py functions are "official" and can 
be used?

> They are also subject to change: for instance LAYER_PAIR does not exist 
> now, due to a collision between 2 structs having the same name defined in
> Kicad code.

True, I can live with that if there is a possibility to savely distinguish 
different API's by version query.

> And this layer pair parameter for EXCELLON_WRITER().BuildHolesList() is not
> so easy to calculate for boards with blind/buried vias.
>
Agreed.

> pcbnew.py is automatically generated by Swig from C++ kicad sources.
> It means many methods in pcbnew.py are not necessary easy to use or not
> necessary useful.
>
> Besides, if you want to work on a tool which needs mainly a list of
> schematic components and their fields, I encourage you to use the netlist
> (.net or .xml files) as entry point, not the schematic files.
>
> Parsing the schematic files is usually tricky, and is justified only to
> access info not in netlist. Especially for complex hierarchies.

The file format of Eeschema is easy to parse, I like it. Using it directly has 
the advantage that users are not responsible to first create an intermediate 
file before using MSEkicadBOM.

> And of course, the code will be outdated when the schematic file format is
> changed.
>
Again, I can live with that if the different formats can be savely 
distinguished.

Martin