scale_x_datetime(): error building plot for early dates

[ASmirnov-HORIS]

Look at the following example:

from datetime import datetime

import pandas as pd

from lets_plot import *
LetsPlot.setup_html()

df = pd.DataFrame(
    {
        'd': pd.to_datetime(
            [
                datetime(1972, 6, 9), 
                datetime(1972, 6, 10), 
                datetime(2021, 3, 5)
            ]
        ), 
        'x': [0, 1, 2]
    }
)

ggplot(df, aes('d', 'x')) + geom_point() + scale_x_datetime()

Here we get:

If we remove the scale, it works:

If we remove the earliest date, it works again:

[aeturrell]

I'm having the same issue with 3.2.0. REPREX:

First the setup:

from lets_plot import *
import pandas as pd

presidential = pd.read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/ggplot2/presidential.csv", index_col=0)
presidential = presidential.astype({"start": "datetime64[ns]", "end": "datetime64[ns]"})
presidential["id"] = 33 + presidential.index

Note that the data are:

	name	start	end	party	id
1	Eisenhower	1953-01-20 00:00:00	1961-01-20 00:00:00	Republican	34
2	Kennedy	1961-01-20 00:00:00	1963-11-22 00:00:00	Democratic	35
3	Johnson	1963-11-22 00:00:00	1969-01-20 00:00:00	Democratic	36
4	Nixon	1969-01-20 00:00:00	1974-08-09 00:00:00	Republican	37
5	Ford	1974-08-09 00:00:00	1977-01-20 00:00:00	Republican	38
6	Carter	1977-01-20 00:00:00	1981-01-20 00:00:00	Democratic	39
7	Reagan	1981-01-20 00:00:00	1989-01-20 00:00:00	Republican	40
8	Bush	1989-01-20 00:00:00	1993-01-20 00:00:00	Republican	41
9	Clinton	1993-01-20 00:00:00	2001-01-20 00:00:00	Democratic	42
10	Bush	2001-01-20 00:00:00	2009-01-20 00:00:00	Republican	43
11	Obama	2009-01-20 00:00:00	2017-01-20 00:00:00	Democratic	44
12	Trump	2017-01-20 00:00:00	2021-01-20 00:00:00	Republican	45

Now the following code, with only the last few entries (shorter time span) included, works:

(
    ggplot(presidential.iloc[-5:], aes(x="start", y="id")) +
    geom_point() +
    geom_segment(aes(xend="end", yend="id")) +
    scale_x_datetime(breaks=presidential["start"], format="%Y")
)

But the full dataset,

(
    ggplot(presidential, aes(x="start", y="id")) +
    geom_point() +
    geom_segment(aes(xend="end", yend="id")) +
    scale_x_datetime(breaks=presidential["start"], format="%Y")
)

Produces an error:

Internal error: IllegalArgumentException : <no message>

[alshan]

Unfortunately, at the moment the earliest date scale datetime can handle is 1.1.1970

@aeturrell, a relatively easy workaround in your case would be to extract year from dates and add "year" as a numeric column to the dataset.
Then you can use regular x/y scales and format year-values using integer formatting "%d".

[alshan]

integer formatting is just "d", sorry.

[aeturrell]

Woop! Thanks so much!