kicad-developers team mailing list archive

[Chris Pavlina <pavlina.chris@xxxxxxxxx>]

On Thu, Sep 15, 2016 at 03:34:03PM -0400, Wayne Stambaugh wrote:
> On 9/15/2016 2:47 PM, Chris Pavlina wrote:
> > On Thu, Sep 15, 2016 at 06:54:59PM +0100, Dan Weatherill wrote:
> >>     Hi all,
> >>     I've been using kicad for about 5 years. Recently (mostly with CERN
> >> changes - especially differential pair support), it's become suitable for
> >> more and more of my more complex projects (I'm switching from Altium
> >> Designer). I also have about 10 years experience working on c++ projects,
> >> but am not what you would call an "expert".
> >>     A feature I've been especially missing in Kicad for several projects is
> >> something akin to Altium's "harnesses": that is, basically a "bus-like"
> >> entity which contains named signals, possibly of different types (e.g.
> >> input, output, passive etc). Harnesses are very useful for drawing things
> >> like e.g. SPI, I2C, CAN connections which have well defined named signals.
> >> Harnesses contain better information than numbered busses for this
> >> application, but can help with schematic "overcrowd" when wiring up chips
> >> with many signals such as micros, FPGAs etc. This is often a big problem
> >> especially in hierarchical designs where to do this using wires requires
> >> possibly dozens of hierarchical labels.
> > 
> > I _really_ want this. IIRC I submitted a wishlist item for it ages ago,
> > and nothing ever really became of it (and as usual I ran out of time to
> > implement it myself). These are hugely useful in complicated designs.
> > 
> >> I've been thinking about implementing this for a while, but have not started
> >> due to being aware that eeschema is under fairly major re-factoring at
> >> present.
> >>
> >>     With guidance from c4758p on #kicad, I think that this feature could be
> >> implemented with the following steps:
> > 
> > For anyone who doesn't know yet, I'm c4757p on #kicad, or c4758p when my
> > connection's dodgy ;)
> 
> So your the guy who's been raking me over the coals on irc. ;)

In my defense, raking people over the coals is basically how IRC works.
Actually I think the protocol doesn't work properly without it. ;)

> 
> > 
> >> 1) add two new items in NETLIST_ITEM_T enum, likely called NET_HARNESS and
> >> NET_HARNESSLABELMEMBER, which have similar functionality to NET_BUS and
> >> NET_BUSLABELMEMBER but for harnesses
> >> 2) several changes to connection rules in netlist.cpp (most majorly in
> >> pointToPointConnect() and a new function
> >> NETLIST_OBJECT_LIST::connectHarnessLabels() )
> >>
> >> 3) minor UI additions (harnesses can be just a different colour wire, and
> >> harness-to-wire connections at this time can be just the same as wire-to-Bus
> >> connections)
> > 
> > In light of the proposal to add line color support to eeschema, I'd like
> > to see something different than just a color. They look like this in
> > Altium:
> > 
> > http://techdocs.altium.com/sites/default/files/wiki_attachments/231555/Harness+Connector+with+entries.png
> > 
> > A point wrt. UI is that when eeschema is converted to GAL, a good bit of
> > the UI code will need to be ported or thrown out. I don't think that
> > should necessarily stall development, but I'd warn you to be prepared
> > and write your code to be properly modular and easy to port. :)
> > 
> >>
> >> The main question usability wise would be how to label harnesses. I see two
> >> possible options:
> >> 1) new syntax: a harness has a label using new type of brackets, with
> >> harness type in brackets, e.g. an I2C bus called A would be "A{I2C}". The
> >> wires would be then labelled exactly like bus labels, e.g.: "A[SCL], A[SDA]"
> >>
> >> 2) bus-like syntax: using same brackets as bus syntax, but make a harness
> >> when the contained type is a string rather than numbers, i.e. "A[I2C]" for
> >> harness label, "A[SCL], A[SDA]" for harness members.
> > 
> > I haven't given this a huge amount of thought yet. Does anybody see a
> > reason for this to me more than just a stylistic detail? We discussed
> > this a bit on IRC and concluded that it would matter if harnesses were a
> > subclass of buses, but if they're a separate object I don't think so.
> > 
> >>
> >>
> >> I guess there are implications for the schematic file system which I need to
> >> understand, and need to be discussed.
> > 
> > As far as I can see, we just need to add a 'harness' type similar to the
> > 'bus' type.
> > 
> >>
> >> All suggestions welcome, and I hope to start working on this soon.
> >> Thanks,
> >> Dan W
> >>
> >>
> >> _______________________________________________
> >> Mailing list: https://launchpad.net/~kicad-developers
> >> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
> >> Unsubscribe : https://launchpad.net/~kicad-developers
> >> More help   : https://help.launchpad.net/ListHelp
> > 
> > _______________________________________________
> > Mailing list: https://launchpad.net/~kicad-developers
> > Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
> > Unsubscribe : https://launchpad.net/~kicad-developers
> > More help   : https://help.launchpad.net/ListHelp
> > 
> 
> _______________________________________________
> Mailing list: https://launchpad.net/~kicad-developers
> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~kicad-developers
> More help   : https://help.launchpad.net/ListHelp