Join on integers, roll = "nearest" and mult = "all" erroneously gives only a single match

Open
#4,382 0 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
3/5
Estimated time
1-2 days
Newbie friendliness
52/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Stale
Tech stack
r
Domain
data

Research direction

Start by running the supplied d1/d2 reproducer with roll = "nearest" and mult = "all", including the constant and differing integer join columns. Trace the rolling-join entry point and its regression tests, then confirm that ties return all expected matches rather than a single row.

Written by the indexing model from the issue text.

Description

non-equi joins

A previous issue described how a rolling join on a column of typeof double, and roll = "nearest" and mult = "all" erroneously resulted in only a single match. A workaround was suggested, where the join column was converted to integer, which then returned the expected number of rows. data.table 1.10.5 IN DEVELOPMENT was used.

However, when trying this workaround with data.table 1.12.9 IN DEVELOPMENT built 2020-02-20 09:26:51 UTC, also the join on integers results in a single match, when several are expected.

An additional curiosity: if the join column in x is constant, the second match is returned. If the join column has different values (both with the same difference to the value in i), the first match is returned. A small example:

The x data, with a constant and a variable integer column:

d1 <- data.table(ri = 1:2, int_const = c(1L, 1L), int_diff = c(1L, 3L))

The i data, with an integer (2L) right in between the values in 'int_diff' above (1L and 3L) :

d2 <- data.table(int = 2L)

(1) Join column in x is a constant integer: only one row from x is returned, the second match:

d1[d2, on = c(int_const = "int"), .(ri, x.int_const), roll = "nearest"]
#       ri x.int_const
# 1:     2           1

(2) Join column in x has two different integers: only one row from x is returned, the first match:

d1[d2, on = c(int_diff = "int"), .(ri, x.int_diff), roll = "nearest"]
#    ri x.int_diff
# 1:  1          1

While digging around, I also tried: list scalar integer in i (.(2L)), other join columns with typeof integer (IDate, ITime) and double (Date). All with same result.

Dominant language
R
Stars
3.9k
Forks
1.1k
Avg merge
14h 4m
Merged PRs (30d)
4

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from Rdatatable/data.table

All issues in Rdatatable/data.table

Similar issues

More R issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.