R1671: Crash on "Undo"

I've switched to release R1671 (Linux). It works very well for me and the  new components are great, but I'm observing one issue:

I can consistently make the application crash ("Segmentation fault") by simply doing an "Undo".

There is a new Undo/Redo system, still unstable.

It would be good to know how to reproduce the crash.

All it takes is to do a right click in the circuit window and select "Undo". That makes the program crash every time on my computer.

I can't reproduce the issue.

Do you mean that doing an Undo always crash no matter what you did before?

Yes, every time.

Yes, this always happens.

I can't reproduce this issue in Linux or Windows.
If you do what I did in this video it crashes?

Windows 11.

It is like there are periods. Sometimes it fails systematically and other times it takes longer. I have observed that saving also previously usually works. In my opinion, in this point, the previous version is more stable.

I have also seen that if a complex scheme is made and executed, ICs fail (for instance counters), but if is tested as it is drawn it works.

I keep testing...

It is like there are periods. Sometimes it fails systematically and other times it takes longer.

We are talking about the application crashing.

When you say "fail" do you mean that it crashes?

Yes. Crashes, I'm sorry.

Here is a demonstration of the "undo crash" on my comouter:



It is reproducable for me.

Ok, thanks.

But if do these exact steps does it crash?
1- Open DiYWaterChiller.sim1
2- Move some component.
3- Undo.

No, undoing moved parts doesn't actually crash the application. It crashes when I undo connections.

Here is a demonstration of the "undo crash" on my computer:

I could reproduce that one.
It is solved at Rev 1677.

No, undoing moved parts doesn't actually crash the application. It crashes when I undo connections.

Nice, then I consider this issue as solved.

This bug is critical, so I will upload new executables.
EDIT: files already uploaded.

Thanks Arcachofo. The 1677 version, it seems that it also solves what I told you about:

"I have also seen that if a complex scheme is made and executed, ICs fail (for instance counters), but if is tested as it is drawn it works."

I also observe that with this Simulide version it goes faster.

Good job!