I'm using the latest version of SM (1.1.8763).
I have auto-calculation enabled.
When I change a variable name, the functions referring to the changed/deleted variable do not show an error due to a missing variable. The functions keep the previous value until I move the function up in the page, to where it is forced to recalculate a different variable.
Selecting "recalculate page" does not help.
See video: recalculate

3DMsFHe.wmv (3,25 MiB) scaricato 706 volta(e).

If you attach the document, it might be possible to doctor.
Jean

I cannot reproduce it in a simplified example, I guess there is something specific that triggers the behaviour shown in the video.

Your file is not an *.sm document.

Wrote

Your file is not an *.sm document.

The video attachment was added by me, since the original is uploaded to an external service and might be deleted.

Wrote

If you attach the document, it might be possible to doctor.
Jean

recalculate.sm (11,83 KiB) scaricato 1794 volta(e).

I've attached a sample.
Instead of deleting the variable, I just clear it.
The last function (highlighted) keeps the old value despite the variable being cleared.
If I drag the clear() to the bottom of the page, the value of the second to last function (above the highlighted one) will not auto-calculate to remove the error. If I click "recalculate", it will remove the error.
If I drag the clear() back up, the second to last function immediately updates to show an error but the last function does not update to show an error.
If I drag the last function up, it will update to show an error.

Wrote

I've attached a sample.

Let me know if it works ... I had a crash on the other PC.
Jean

recalculate.sm (11,85 KiB) scaricato 1729 volta(e).

Aah, I get it now. This should be a bug.
While Clear(F.excitation) is enabled F.T can be calculated, obviously.
But when Clear(F.excitation) is disabled once F.T is calculated, F.T value remains.
Even there is no F.excitation because of Clear().

Changing all optimizations to symbolic doesn't matter too.
I will say it again, this should be a bug.