Bug- 4.1.2 (and 4.1.1) report "bad connections" unnecessary.

[MatNieuw]

Mint 22.3 XFCE desktop, offline version of 4.1.2js.

When opening a large circuit with active scope and lots of subcircuits, "bad connections" are reported for open lines. If you attach an open wire to any of those, all disappear. This is a fairly recent bug, did not happen in the past. Or maybe the increasing complexity of the circuit caused it, it has nested subcircuits.

[pfalstad]

Does it happen without running=false?

[MatNieuw]

Yes. If I use the online version, it also reports bad connections when I load the circuit, whether I stop it running before loading the circuit or not. See screenshot. I should note I always have option small grid for the offline version and no toolbar, but with the online version it happens with the default values.

[pfalstad]

Hmm. Can you send the circuit?

[MatNieuw]

It is the same as the circuit I sent for issue #360. I just discovered a problem with a subcircuit in the big ctrl-7 subcircuit, that affects the logic for me, but should have no effect on the sim itself. The problem is because of switching between 4.1.1 and 4.1.2, I now only use 4.1.2 .

[pfalstad]

I don't see any circuit in the issue #360 listing. Did you email it to me? I didn't get it.

[MatNieuw]

Sorry, was sent to the wrong address, you should have it now. I just noticed that when I want to edit the big sub circuit CTRL-7 seperately , it reports "singular matrix", so obviously there is an error in it. Maybe that has to do with the "bad connections" .

[pfalstad]

the "singular matrix" seems like a bug, because it works fine unless you try to edit it. also there are "missing node" errors reported in the console.

[pfalstad]

The issue is detection of bad connections right after loading the circuit. The bounding box of subcircuits/chips is bad when first loaded from a circuit, depending on how you created it (it saves the location you drag to, as if it were a 2-terminal device). As soon as you modify the circuit in any way it fixes everything.

Example:

https://www.falstad.com/circuit/circuitjs.html?ctz=DwYwlgTgBAZgvAIgIwKgFwM6IAwDpsEECsqYIiSeATAVQOx0DM2AHFQGwCcndqIARoiLZUAB0EJhqAG4QhqALaYhAUwC0SFAD4AUFCigQUAB6JGjdlDUAWFlEpVrUa53ap4yRQHtEAExUwAIYArgA2aAgA9Lr6wNCmknSWSNZO7Kz2qe44qHLIhFExBvFC6VA0dul2FdkIIlB5lASFesUmiEzJVHad9pxUtfWNBdGtce0IdNhOVESWSdVzg7kUI7rAkeAQukA

[pfalstad]

The "exception in stampCircuit" is because you have some labeled nodes (like F-T_out) that are not connected to anything. Another bug.

[pfalstad]

I see "exception in stampCircuit" when I try to view (not edit) the CTRL-7. Are you also seeing a "singular matrix" somewhere?

[MatNieuw]

The F-T_out is normally connected, I just removed the logic block at the cross of line 6 with P3 for a while.

I fixed the logic error in CTL-7, but it did not make a difference for the rest.

The "singular matrix" msg I didn't see when viewing CTRL-7, but it is there when editing it from the source file. But it was fixed by changing all of the logic blocks (really specially wired single relays), which were wrong, a leftover from 4.1.1 , and renaming it CTRL-8 . I did not see the "exception in StampCircuit" in CTRL-7 (not now either), but I don't have the dev tools visible most of the time. I'll do that more often.

The "bad connections" indeed go away when changing something, even moving a text a little. Thanks.

Ah, the bounding box depends on the first location. I typically draw a subcircuit somewhere in a free area where there is enough free space, then move it to the right position.

I'll retest tomorrow with 4.1.3alpha, thanks.

[pfalstad]

"exception in StampCircuit" was/is shown at the bottom of the window for me on 4.1.2.

if you move the subcircuit it moves the bounding box, I think. but it depends on how far you drag the mouse when you create it, that determines it size. this is fixed in 4.1.3alpha