On Jul 19, 2017, at 10:22 AM, Ouabache Designworks (z3qmtr45 AT gmail DOT com) = [via geda-user AT delorie DOT com] <geda-user AT delorie DOT com> wrote:

On Wed, Jul 19, 2017 at 5:17 AM, Nicklas Karlsson = (nicklas DOT karlsson17 AT gmail DOT com) [via geda-user AT delorie DOT com] <geda-user AT delorie DOT com> wrote:

gschem is actually rather versatile. It works: circuit boards, cable = drawings, VHDL, linuxcnc configuration, mathematica, tango, ... and = should work whenever there is a need to draw a schematic with symbols = there pins are connected together with lines, for example a hydraulic = circuit. Is is possible assemble several signals into a bus which is = like a cable with several strands.

What attracted me = to Cschem was the promise that it would be workflow agnostic. The = problem with gschem,kicad and about all other schematic capture tools is = that they are all designed to be PCB front end tools. You can have any = feature as long as it would be useful in a PCB tool flow.

We need a tool that is = targeted to the embedded system architect that also supports state = machines, flow charts etc. A tool for the architects would be a superset = of all other engineering tools and would satisfy everyones needs. Make = it modular so that it doesn't become bloated.

But that kind of flexibility is gschem=E2=80=99s = strength. By coincidence, my work today involves drawing system diagrams = in gschem (OK, actually lepton-schematic), exporting the connectivity to = Mathematica (gnetlist -g mathematica). I then crunch those algebraicly = through Laplace transforms and Mathematica=E2=80=99s system modeling = stuff into discrete-time approximations suitable for CUDA = implementation,

From: John Doty <jpd AT noqsi DOT com>
Subject: Graphical Makefiles
Date: October 12, 2015 at 11:52:43 AM MDT
To: geda-user AT delorie DOT com

On Aug 25, 2015, at 4:23 PM, = Evan Foss (evanfoss AT gmail DOT com) [via geda-user AT delorie DOT com] <geda-user AT delorie DOT com> wrote:

I almost want to write a = netlister that lets you move files and the
functions that = process them around in a schematic so that you have a
graphical view of workflow.
gschem -> = schematic file -> gnetlist -> makefile

Try this:

We need = to get that back with Cschem. Open source is the only way to do it = because a engineer can take their toolkit with them as they move between = companies.

It is a little bit hard to figure out what is wrong/missing in = gschem:
+ Something make movement of symbols with connections a little = cumbersome.

The PCB world has a lot of great tools = for moving parts on a board with a ratsnest, auto placing parts and = autorouting. I would love to bring these into a schematic capture tool. = Autorouting is a lot easier if you can cross lines and not make a = connection unless you put a dot on it.

+ Then connecting from a bus it should be possible to select which = signal should be used.
+ netlist so that available net could be listed and assigned = attributes, not just net segments.
+ It should be possible to see names of nets in busses.

gschem doesn't really have true busses. It has scalers and = vectors. True busses would let you do all of that. IC design tools can = do it but they never made it into the PCB world

- Mapping of pins to footprints should be a plugin or backend, not = all circuits need this.
+- Available standard attributes would come from plugin or backend = selected from symbol.
0 Choice between embed/link is already available.
+ bus pins.
+ backend may provide online design rule check.
? backend need to be available on context, different symbols have = different backends to make list of available attributes or?
=

I came to the conclusion gschem is very versatile with quite little = missing but something is missing. To add support for different = clearances between net a backend would be needed to specify clearances = between nets but this only make sense for physical electric circuits. = Support to select track width or cable dimension could could be done = with the same approach. a netclass or track width attribute could be = added to net as is today but it might be better to point the other = direction and list the net names?

Yes, Use Cschem for entry, editing and = viewing. Use separate backend tools to o the real work.

John Eaton