Closed awb99 closed 1 year ago
:date | :open | :high | :low | :volume | :close |
---|---|---|---|---|---|
2023-09-01T00:00 | 137.49 | 137.5000 | 134.8500 | 16185949 | 135.66 |
2023-09-05T00:00 | 135.37 | 136.4200 | 134.5801 | 14948394 | 135.75 |
2023-09-06T00:00 | 136.02 | 136.5299 | 133.6650 | 14601212 | 134.46 |
2023-09-07T00:00 | 133.57 | 135.5800 | 132.9500 | 14062620 | 135.27 |
2023-09-04T00:00 | 135.27 | ||||
2023-09-08T00:00 | 135.27 | ||||
2023-09-11T00:00 | 135.27 | ||||
2023-09-12T00:00 | 135.27 | ||||
2023-09-13T00:00 | 135.27 |
The above table shows September 4 (which does not have a bar) gets added below. So all rows that do not have the row in the right dataset are in the end of the dataset.
I'll take a look at that today. One question: which version of TMD is on your classpath?
left-join
on single column delegates action to tech.v3.dataset.join/left-join
so I think it is a problem on TMD side.
Ok, verified in the TMD source code. It looks like expected behaviour. Nothing to do here.
When I call tablecloth left-join then it adds rows that are not in the right dataset into the end. I guess this is a bug in tablecloth. Because left-join should not really affect the sort order.
My example: calendar: dataset sorted by :date (it only has :date column) ds-bars: dataset sorted by :date (columns: :open :high :low :close :volume :date)