X-Authentication-Warning: delorie.com: mail set sender to geda-user-bounces using -f X-Recipient: geda-user AT delorie DOT com X-TCPREMOTEIP: 173.48.170.224 X-Authenticated-UID: jpd AT noqsi DOT com From: John Doty Content-Type: multipart/alternative; boundary="Apple-Mail=_731D5C8D-1345-449D-B234-0CFD00B0E660" Mime-Version: 1.0 (Mac OS X Mail 10.3 \(3273\)) Subject: Re: [geda-user] Refdes bug or Master Attribute Document on the Wiki needs update. Date: Wed, 30 Jan 2019 18:21:02 -0500 References: <9ed059c0-f3c5-1482-169b-f8f1119f3208 AT fastmail DOT com> <5BC4365D-FBD0-4495-806B-C30BA710D31B AT noqsi DOT com> <4739A346-6DB4-4250-965C-13AA19863969 AT noqsi DOT com>

To: geda-user AT delorie DOT com In-Reply-To: Message-Id: <63ADD01D-C463-4478-B7C1-01377D67095D@noqsi.com> X-Mailer: Apple Mail (2.3273) Reply-To: geda-user AT delorie DOT com Errors-To: nobody AT delorie DOT com X-Mailing-List: geda-user AT delorie DOT com X-Unsubscribes-To: listserv AT delorie DOT com Precedence: bulk --Apple-Mail=_731D5C8D-1345-449D-B234-0CFD00B0E660 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=utf-8 > On Jan 30, 2019, at 3:30 PM, Roland Lutz wrote: >=20 > On Wed, 30 Jan 2019, John Doty wrote: >> Can you test netlist topology for equivalence? >=20 > Graph isomorphy is not a trivial problem, but if you leave the case of = mass-renaming packages and nets out, it should be implementable without = too much of a problem. As I=E2=80=99m moving my infrastructure over to Lepton, one of the = things on my list is a plugin to implement stable naming for unnamed = nets. The idea is that (refdes pinnumber) pairs define a unique = connection, so choose an ordering, sort them, and use the first one as = the net name. So, unnamed_net27 might become J1_17. That=E2=80=99s handy = for a number of things like SPICE simulations, and it would make = comparing topologies easier. >=20 > Do you have a specific use case in mind? >=20 >=20 One thing I=E2=80=99ve done is extract the netlist from an Osmond layout = and compared it to an Osmond netlist created by gnetlist to verify that = nothing got changed in layout. That one=E2=80=99s not hard: the layout = file contains a netlist with the names of things unchanged, only with = vias added. It=E2=80=99s easy enough to remove the vias, sort the = remaining connections, sort the nets, and compare to a similarly = processed netlist from gnetlist, thus verifying that the layout reflects = the schematics. I=E2=80=99ve even used this approach to recover a = schematic from a layout that was fairly drastically revised (replacement = of a microcontroller with an FPGA!) without much trouble. John Doty Noqsi Aerospace, Ltd. jpd AT noqsi DOT com --Apple-Mail=_731D5C8D-1345-449D-B234-0CFD00B0E660 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=utf-8

On Jan 30, 2019, at 3:30 PM, Roland Lutz <rlutz AT hedmen DOT org> = wrote:

On Wed, 30 Jan 2019, John Doty wrote:
Can you test netlist topology for = equivalence?

Graph isomorphy = is not a trivial problem, but if you leave the case of mass-renaming = packages and nets out, it should be implementable without too much of a = problem.

As I=E2=80=99m moving my infrastructure over to Lepton, = one of the things on my list is a plugin to implement stable naming for = unnamed nets. The idea is that (refdes pinnumber) pairs define a unique = connection, so choose an ordering, sort them, and use the first one as = the net name. So, unnamed_net27 might become J1_17. That=E2=80=99s handy = for a number of things like SPICE simulations, and it would make = comparing topologies easier.

Do= you have a specific use case in mind?

One thing = I=E2=80=99ve done is extract the netlist from an Osmond layout and = compared it to an Osmond netlist created by gnetlist to verify that = nothing got changed in layout. That one=E2=80=99s not hard: the layout = file contains a netlist with the names of things unchanged, only with = vias added. It=E2=80=99s easy enough to remove the vias, sort the = remaining connections, sort the nets, and compare to a similarly = processed netlist from gnetlist, thus verifying that the layout reflects = the schematics. I=E2=80=99ve even used this approach to recover a = schematic from a layout that was fairly drastically revised (replacement = of a microcontroller with an FPGA!) without much trouble.

John Doty = Noqsi = Aerospace, Ltd.

jpd AT noqsi DOT com

= --Apple-Mail=_731D5C8D-1345-449D-B234-0CFD00B0E660--