[Fixed] Bug when inserting a row into spreadsheet

[wang02139]

When a spreadsheet has entries with aliases which goes through 2 levels of reference, and the user inserts a row, the reference is broken and there is an error that a property is not found.

I have attached a very simple file with this problem. If you insert a row at, say, row 3, there is an error

"15:19:52 <Spreadsheet> Cell.cpp(594): Unnamed#Spreadsheet.A9: Property 'b' not found in 'b'"

There are 2 aliases:
a = 1
b = a + 1
The last cell is b+1 which has the error.


OS: Ubuntu 18.04.5 LTS (ubuntu:GNOME/ubuntu)
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.19.22284 (Git) AppImage
Build type: Release
Branch: master
Hash: bf1e8e48389f5e9e25bd77b67fe98da4213e797c
Python version: 3.8.5
Qt version: 5.12.5
Coin version: 4.0.0
OCC version: 7.4.0
Locale: English/United States (en_US)

[chrisb]

This was reported recently.

[wang02139]

Thanks. Can you point me to the discussion in the forum or bug tracker?

[TheMarkster]

Refreshing fixes it for me.

[wang02139]

We still want to file and fix the bug, right?

[TheMarkster]

I looked through the source code, but I can't figure out what the problem is. I can call a recompute to fix it and suppress the exception message, but that's really more of a hack than an actual fix. I'm not sure if it's worth submitting as a PR. Thoughts?

[wang02139]

This bug is quite bad. I have a large spreadsheet with lots of parameters. When I insert a row, I was badly surprised that many cells have errors like this. Most users wouldn't know to refresh to work around this bug.

I also don't think recompute is the right way to fix it. Maybe the cells should be updated in a correct order. Let's file a bug report first. I'll also look at the spreadsheet code to see how to fix it.

[wang02139]

https://tracker.freecadweb.org/view.php?id=4429

[chrisb]

Searching for "Spreadsheet insert row bug" shows this as the first(!) result: https://forum.freecadweb.org/viewtopic.php?t=49969. The usual procedure here is to prepare things in a discussion as good as possible before creating a ticket.

[wang02139]

OK. I'll be more careful next time.

I looked at the code and understood the problem now:

Here it's using revAliasProp to look for the location, but it's the old location of the alias:
https://github.com/FreeCAD/FreeCAD/blob ... t.cpp#L993

The alias may be changed later in moveCell which updates revAliasProp
https://github.com/FreeCAD/FreeCAD/blob ... t.cpp#L708

Thus the dependency graph is wrong. It's not an easy fix. I'll think of a clean way to fix it.