legend shows additional information not in data set

[Alcampopiano]

Hi all,

Why does 2011 show up in the legend if there are no records for 2011 in the DataFrame?

Perhaps I am missing something simple...

weather = data.seattle_weather()

alt.Chart(weather).mark_line().encode(
    x='month(date):N',
    y='max(wind)',
    color='year(date):N'
)

[jakevdp]

This is due to an issue with timezones that has been fixed in the master (See #1027 and #1053).

In short, the input is being treated as UTC, but rendered in your local time, so there is a timezone correction that shifts 2012-01-01T00:00 back into the year 2011.

I'm currently writing documentation to summarize how to deal with this kind of issue (PR shortly), and it will be part of the 2.2 release that should come in the next couple days.

[Alcampopiano]

There is always a logical explanation! Thanks Jake.

Is the 2.2 version leveraging vega-lite 3?

[jakevdp]

No, we're still catching up with VL 2.6

We'll plan to support VL 3 in the 2.3 release.

[jakevdp]

(Vega-Lite 3 is still only a release candidate at this point)

[Alcampopiano]

Okay great. Thanks very much.

[jakevdp]

timezone doc update is here: #1087