Non-equi join results in misleading variable names

Hi,

I've run into the following unexpected issue...

I am trying to join sales to shifts here by timestamp.

sales = data.table( sale_id = 1:5, sale_ts = (1:5)*5 )
shifts = data.table( shift_id= 21:23, shift_start = c(1,10,15), shift_end=c(9,14,19) )

sales[shifts, on=.(sale_ts>=shift_start, sale_ts<shift_end) ]

returns

   sale_id sale_ts sale_ts.1 shift_id
1:       1       1         9       21
2:       2      10        14       22
3:       3      15        19       23

This is misleading because the timestamps that are returned are not sale timestamps at all, but shift timestamps. I realize that reversing the code to do shifts[sales ... will give more appropriate results, but I'm likely not the only one confused by the behavior of the former formulation and possibly not the only one for whom this behavior resulted in some headache.

Thanks!

Seems related to join column naming with on syntax

Yes, essentially the same issue as #1807 and a couple others. I hadn't done a good enough job searching. Closing as duplicate.

I do think since base merge doesn't support non-equi joins, it made sense there, but I think the SQL naming conventions make more sense for data.table.

Navigation Menu

Repository navigation

Non-equi join results in misleading variable names #2307

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Non-equi join results in misleading variable names #2307

Description

Activity

Henrik-P commented on Aug 17, 2017

aleks-via commented on Aug 17, 2017

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions