Unify recursive update on ReplacementPane

[Hoxbro]

Fixes #4951
Fixes #4968

[codecov]

Codecov Report

Merging #4958 (f97d028) into main (2ab34b3) will decrease coverage by 10.05%.
The diff coverage is 87.50%.

@@             Coverage Diff             @@
##             main    #4958       +/-   ##
===========================================
- Coverage   83.49%   73.44%   -10.05%     
===========================================
  Files         270      271        +1     
  Lines       38239    38281       +42     
===========================================
- Hits        31926    28116     -3810     
- Misses       6313    10165     +3852     

Flag	Coverage Δ
ui-tests	?
unitexamples-tests	73.44% <87.50%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
panel/pane/base.py	86.17% <73.33%> (-0.92%)	⬇️
panel/tests/test_param.py	99.71% <100.00%> (+<0.01%)	⬆️

... and 57 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[Hoxbro]

I think it would be beneficial to have _generate_hash as a fallback and do the simple equally check first. Mainly because it _generate_hash is significantly slower for most cases.

Maybe even have a .all() try before using it.

[philippjfr]

My thinking was that in most cases the cost is neglible but in the case of a dataframe, where the cost does potentially matter significantly, simple equality won't, so you end up paying the cost twice.

[philippjfr]

My thinking was that in most cases the cost is neglible but in the case of a dataframe where the cost does potentially matter significantly equal won't work anyway so you end up paying the cost twice.

The .all() could help with that. Not sure I prefer it though.

[Hoxbro]

My thought was that we should not penalize every other type because Arrays and DataFrames are ambiguous with more than one element.

The .all is also faster than having to generate the hash.

[philippjfr]

Fair points, swapped the order.