← Back to team overview

kicad-developers team mailing list archive

Re: [PATCH] eeschema: invisible pin connection

 

Yes I understand, but proposing that changing the code for everyone using it to something more confusing isnt really a good solution to me.
Its a work-around that will be defacto standard.

Alternatives to changing the netlist generation:

- Remove all invisible pins altogether

- Make all no-connect pins unconnectable

- Make invisible pins that are connected visible somehow that doesnt break pin stacking, maybe making them grey as in the symbol editor. still warn with ERC.

- Create logic to properly handle internal connections in footprints.

All of theese are better choices in my opinion than adapting the code to fitting the dumbass library designer.


On 2017-02-07 13:49, Chris Pavlina wrote:
You do realize that symbols are made and used by different people,
right? The person placing a symbol in the schematic DOESN'T KNOW that
the dumbass library designer put a hidden pin in it. They just wonder
why ERC is complaining about a connection somewhere where there is no
pin (that they can see).

On Tue, Feb 07, 2017 at 01:46:17PM +0100, Kristoffer Ödmark wrote:
I think having a pins function changing depending on its relative position
to other pins is more confusing, especially if it is toggled by a checkbox
saying "visible".

In that case a some kind of indication is better. Not changing the
connectivity of the pin by hardcoded logic.

It seems to me this is still an issue that can be fixed by the ERC or
checking manually. Not using the ERC is also bad practice, and reworking
this way is just enforcing two bad behaviours.

To me every no-connect pin should not be able to be connected. Not depending
on its visibility.

- Kristoffer

On 2017-02-07 13:22, Chris Pavlina wrote:
On Tue, Feb 07, 2017 at 01:15:43PM +0100, Kristoffer Ödmark wrote:
I think the aim then should be to inform about this. I see the "invisible"
checkbox as being just that, it makes the pins invisible, but still
connected.

Shouldnt this be a warning issue for the ERC, connecting to an invisible pin
that is not stacked?

And as you said, you had to clean out parts that had invisible pins that
that was supposed not to be connected. Fault of creating the symbol, I think
the symbol should be reworked instead of hardcoding around faulty symbols.

There are many silly ways of using stuff, I dont agree that having a
visibility checkbox determining if it is connectable is the right way,
rather have a pop-up warning that says that you have connected to an
invisible pin.

...you don't think kicad has enough popup warnings /yet/?! Are you
kidding?

The feature is confusing, it should be reworked to be less confusing.
Not leave it confusing and yell at the user when he gets confused.


- Kristoffer

On 2017-02-07 12:50, Chris Pavlina wrote:
On Tue, Feb 07, 2017 at 12:44:45PM +0100, Kristoffer Ödmark wrote:
I wasnt saying its a good idea, but having invisible pins indicates that you
want to connect to something that is not visible, its literaly there in the
name. An invisible pin.

I have seen numerous parts made by people who clearly don't get that, as
they think invisible pins are a nice way to represent no-connect pins
visibly in libedit that don't show up in and clutter the schematic. Just
had to clean a bunch of those out of my own library that someone
submitted, and someone else said the official KiCad libs have a bunch
too. Not sure why you think it's so obvious when actual usage shows it's
not.


I mean, otherwise there could be a stacked pin instead. Im not saying that
invisible pins are good practise, but thats not really for me to say.

What is silly is having invisible pins working as no-pins except if they are
a stacked pin, well that doesnt sound clear to me.

What's silly is using them that way when you could just hide the pin
text. It's only the text that collides and makes them look bad.

Compromise: don't connect invisible pins of type "no connect". Remove
the stupidity in the design without screwing the people who depended on
it.


-Kristoffer

On 2017-02-07 12:33, Chris Pavlina wrote:
Honestly I think that's one of the silliest things I've ever heard. Pins
that you can't see should make connections that you can't see to wires
that you can? The ONLY imaginable use case for this is stacks of pins.
Every other possible case is a mistake.

On Tue, Feb 07, 2017 at 11:09:44AM +0100, Kristoffer Ödmark wrote:
Honestly I think the invisible pins are supposed to work exactly as they
are, that they should be able to connect, otherwise there are the "no
connect" - pin type or the option of just removing the pin from the symbol
altogether.

- Kristoffer


On 02/07/2017 10:02 AM, Oliver Walters wrote:
Kristoffer this is good feedback. I did not expect this to get pushed
straight away, and perhaps there is a way forward that won't break
schematics.

Relying on implicit connected that is *not* displayed on the schematic
seems like a very bad idea to me.

I appreciate your use case (I currently have a few symbols that do that
too).

On Tue, Feb 7, 2017 at 7:57 PM, Kristoffer Ödmark
<kristofferodmark90@xxxxxxxxx <mailto:kristofferodmark90@xxxxxxxxx>> wrote:

This seems dangerous, I have seen a few design where there are 5-10
pins hidden under the same pin, excpecting them to be connected.

I would rather this hidden connections were indicated in some way,
this change disconnects lines and might break some users
footprints-symbols connection.

- Kristoffer


On 02/07/2017 09:47 AM, Oliver Walters wrote:

    Hi all,

    The attached patch prevents invisible pins from being connected
    using
    the wire tool in eeschema.

    a) If you connect a wire endpoint to the same position as a pin
    endpoint, they are NOT connected visually
    b) Wires and insivible pins are also ignored during netlist creation
    c) This does not affect the ability of invisible power-pins to
    automatically connect to global power labels

    Is the current behavior of connecting invisible pins to wire
    endpoints
    desired? Or is it just an aberration?

    If there is a very good reason that pins not visible in the
    schematic
    are able to be connected silently?

    before: http://i.imgur.com/3gModvW.png

    after: http://i.imgur.com/r8O7c3Y.png

    (Note the 'dangling' wire-end indication)

    Cheers,
    Oliver




    _______________________________________________
    Mailing list: https://launchpad.net/~kicad-developers
    <https://launchpad.net/~kicad-developers>
    Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
    <mailto:kicad-developers@xxxxxxxxxxxxxxxxxxx>
    Unsubscribe : https://launchpad.net/~kicad-developers
    <https://launchpad.net/~kicad-developers>
    More help   : https://help.launchpad.net/ListHelp
    <https://help.launchpad.net/ListHelp>


--
 -Kristoffer

_______________________________________________
Mailing list: https://launchpad.net/~kicad-developers
<https://launchpad.net/~kicad-developers>
Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
<mailto:kicad-developers@xxxxxxxxxxxxxxxxxxx>
Unsubscribe : https://launchpad.net/~kicad-developers
<https://launchpad.net/~kicad-developers>
More help   : https://help.launchpad.net/ListHelp
<https://help.launchpad.net/ListHelp>



--
-Kristoffer

_______________________________________________
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


References