Subset by bug

[MarkusBonsch]

Closes #2713 and #2591:

I fixed a bug in src/dogroups.c that caused wrong entries in the grouping column if:

• by was used during a join
• on= was used to specify the join columns
• The grouping column was sorted and contained groups longer than 1 row.

The following example from issue #2713 works correctly now:

library(data.table)
DT = data.table(
    id = c("a","a","a","b","b","c","c","d","d"),
    group = c(1,1,1,1,1,2,2,2,2),
    num = 1)
#    group V1
# 1:  1     2
# 2:  2     2

I updated tests and added a comment that explains the reasoning for my change into the source code.

[codecov-io]

Codecov Report

Merging #2748 into master will increase coverage by 0.09%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #2748      +/-   ##
==========================================
+ Coverage   93.43%   93.53%   +0.09%     
==========================================
  Files          61       61              
  Lines       12296    12495     +199     
==========================================
+ Hits        11489    11687     +198     
- Misses        807      808       +1

Impacted Files	Coverage Δ
src/dogroups.c	93.35% <100%> (-0.05%)	⬇️
src/fread.c	98.19% <0%> (+0.21%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c579e01...95b3692. Read the comment docs.

[mattdowle]

Fantastic! Thanks for quick fix.