delorie.com/archives/browse.cgi | search |
On Tue, 12 Jan 2016, Peter Clifton (petercjclifton AT googlemail DOT com) [via geda-user AT delorie DOT com] wrote: > > > On 12 Jan 2016 17:02, "Frank Miles (fpm AT u DOT washington DOT edu) [via geda-user AT delorie DOT com]" <geda-user AT delorie DOT com> wrote: > > > > > Perhaps I'm confused. I thought I had read that the netlist import was > > deprecated, and that the proper way of importing the netlist was through > > PCB:File/Import Schematics. If netlist import is still equally valid, > > then please accept my apology for the needless noise. > > > > -F > > The import mechanism was added as a more integrated way of retrieving this information, rather than relying on an external tool, > gsch2pcb to process your existing design file and make edits. (This tool had to read and make edits to the board files, so > represented a barrier to format changes. This was probably not the initial motivation of the new mechanism, but represents a big > plus point to its existence). > > Underlying the mechanism is a netlist backend for gnetlist which describes the components and nets of the design using a series of > pcb action commands. These are executed by pcb on the open design, and in turn update the internal view of the netlist. > > The "old" netlist format is not deprecated, although it does not convey all of the information about parts etc.. (gsch2pcb managed > that aspect based upon a template file a separate gnetlist backend would populate with all the parts of a given design). > > Since gsch2pcb and associated netlist backends were in a separate code base to pcb, this represented a compatibility problem if ever > we changed things with the file format. The new backend lives with PCB (iirc), so merely relies on gnetlist keeping a stable scheme > api and command line interface. > > You might be interested to know that the update mechanism can be wired to call a makefile (or other scripts?) to generate the > required data, so it is infact quite general. Only the "schematics" attributes are really assumed to come from a simple gschem work > flow. > > At some point, I may update xgsch2pcb to use this new mechanism, rather than gsch2pcb. It seems redundant to update gsch2pcb itself > though. > > Peter Thanks to you - Peter, Nicklas, and DJ - for the explanations. One more question - where can I read about the differences, and how this update mechanism might be used? -F
webmaster | delorie software privacy |
Copyright © 2019 by DJ Delorie | Updated Jul 2019 |