Improve clarity of pandas dataframe refresh tooltip

[mattrunyon]

Description

Originally I filed this because I thought the button should refresh a dataframe created from a ticking table.

The tooltip to refresh a dataframe is unclear. Since Deephaven is based around ticking data, the tooltip as is can easily be misunderstood to mean it should update as if the dataframe ticked.

The engine takes a snapshot when creating a dataframe from a ticking table. That snapshot never updates. The refresh button is only applicable if you mutate the dataframe.

[dsmmcken]

I don't think that is expected to work. My understanding is to_pandas is a one way a static snapshot at the time of creation. The refresh is there for any subsequent pandas operations.

[mattrunyon]

The tooltip is too ambiguous then. It just says "Click to refresh pandas dataframe. Updates do not occur automatically"

Is the button just to update if the user performs some mutating pandas operation then? If it's non-mutating then the REPL would open the new version created automatically

[dsmmcken]

Correct, it's for mutations. to_pandas doesn't create a magically live pandas dataframe.

[mattrunyon]

Thoughts on clarifying the tooltip? Based on the tooltip I thought it would be for ticking data updates. Makes sense from the engine standpoint that it's just a snapshot, but from a user using Deephaven, I defaulted to thinking about ticking data as the likely meaning.

"Click to refresh dataframe. Pandas mutations do not display automatically" or something like that? From the pandas docs, they are mutable, but they say most operations do not mutate the original dataframe. See here

[dsmmcken]

Sure. Basically if it doesn't re-assign a new dataframe it won't update. Though I would argue most of your confusion here was with the to_pandas method, and you should make sure that docstring made sense to you.

Run:

import pandas as pd

data = {'Name': ['Jai', 'Princi', 'Gaurav', 'Anuj'],
        'Height': [5.1, 6.2, 5.1, 5.2],
        'Qualification': ['Msc', 'MA', 'Msc', 'Msc']}

df = pd.DataFrame(data)

Then: a. won't update:

  address = ['Delhi', 'Bangalore', 'Chennai', 'Patna']
  df['Address'] = address

b. would update:

  df = df.assign(address=['Delhi', 'Bangalore', 'Chennai', 'Patna'])

[vbabich]

Change text to "Changes to dataframe don't appear automatically. Click to refresh."

[dsmmcken]

Thought about, and I am not convinced this wording is actually better. "mutations" is too complex of a term, "changes" is no better than "updates". I am going to close this.