Regression in Tabulator widget for panel>0.12.1

Copying selected rows from the DataFrame behind a filtered Tabulator widget succeeds for panel=0.12.1 but fails with an IndexError for panel>=0.12.2.

Environment

OS:

• Windows 10
• version 1909
• build 18363.1679

conda environment:

• bokeh=2.4.1
• ipython=7.29.0
• jupyter_client=7.0.6
• jupyter_core=4.9.1
• notebook=6.4.5
• openpyxl=3.0.9
• pandas=1.3.4
• panel=0.12.4 (or 0.12.3 or 0.12.2)
• param=1.12.0
• python=3.9.7

Steps to reproduce

1. Create conda environments from the .yml files.
2. Run the MRE in the Jupyter notebook Tabulator-regression.ipynb
3. The bug can be reproduced by either:
   • running the bottom 4 cells of the notebook; or
   • using the GUI:
     • enter e.g. "cut" in the "descr" filter box;
     • select some or all of the filtered rows;
     • click the "Add selected rows" button;
4. Result:
   • 5 rows appear in the lower Tabulator (for panel=0.12.1)
     • top 2 rows selected from the unfiltered (top) Tabulator
     • bottom 3 rows selected from the filtered (top) Tabulator
   • 2 rows appear in the lower Tabulator (for panel>=0.12.2)
     • only the top 2 rows selected from the unfiltered (top) Tabulator
     • no rows are added when selected from the filtered (top) Tabulator

Results

Expected result

Running all the cells in the notebook with panel=0.12.1 will show the following in the lower Tabulator:

code	descr	temp
00	Under the Weather	1.0
01	Top Drawer	1.0
03	Cut To The Chase	1.0
07	A Cut Above	1.0
08	Cut The Mustard	1.0

Actual result

Running all the cells in the notebook with panel>=0.12.2 will show the following in the lower Tabulator:

code	descr	temp
00	Under the Weather	1.0
01	Top Drawer	1.0

Screenshots

Succeeds for panel=0.12.1

Fails for panel>=0.12.2