[FR] Support timestamp input for timeseries

[Tyrannas]

Apparently right now xan does not support timestamps collumns for ploting time series:

xan plot: could not parse "789174000" as date!

The same value works when being parsed by python data.fromtimestamp() method

[Yomguithereal]

So, I could support timestamps but my issue is the following: There is no way to infer the granularity of a timestamp. That is to say it could be a traditional "seconds" timestamp, but it could also be a floating point one for milliseconds support, or even a integer one supporting milliseconds. So I don't really know what I should do here. I could add flags letting the user state that input data are timestamps (and what kind of them they are).

Now, a workaround for the time being is to use xan map, xan select -e or xan transform ahead of the xan plot command to parse the timestamp using moonblade functions (timestamp, timestamp_ms and datetime with a format argument, respectively).

[Tyrannas]

Hum yeah didnt think of the multiple timestamps formats but doesnt the format depend on the length of the number provided ? But yeah I guess using the xan transform should be enough, closing it with an example for reference:

xan transform epoch "timestamp(epoch)" csv_output.csv | xan plot -LT epoch --count

[Yomguithereal]

Hum yeah didnt think of the multiple timestamps formats but doesnt the format depend on the length of the number provided ?

Nope, everything goes out of the window as soon as you approach 1970 (to distinguish between s & ms I mean).

xan transform epoch "timestamp(epoch)" csv_output.csv | xan plot -LT epoch --count

Just a little tip about transform, you can use _ as a placeholder to avoid repeating epoch:

xan transform epoch "timestamp(_)" csv_output.csv | xan plot -LT epoch --count

I will reopen this though because I believe we should be able to do something to make this easier.

[Tyrannas]

Oh this shortcut is actually really helpful