Add skip_absent to setcolorder

[sluga]

Motivation: I sometimes want to ensure a certain order of columns but the set of columns is not fixed, i.e. a given column may or may not be present.

Now, working around that is trivial & so the benefit of this new argument isn't large, but it would be a small quality of life improvement & furthermore setnames already has the skip_absent argument, so it feels appropriate to add it to setcolorder as well.

[MichaelChirico]

please revert whitespace changes first to make the diff clearer

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.53%. Comparing base (8476b5b) to head (b667257).
Report is 24 commits behind head on master.

❗ Current head b667257 differs from pull request most recent head b135a29. Consider uploading reports for the commit b135a29 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #6044      +/-   ##
==========================================
+ Coverage   97.51%   97.53%   +0.01%     
==========================================
  Files          80       80              
  Lines       14980    14922      -58     
==========================================
- Hits        14608    14554      -54     
+ Misses        372      368       -4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[MichaelChirico]

Hi @sluga, happy to review but I still see a lot of spurious whitespace diff in R/data.table.R, PTAL

[sluga]

Missed the indent change. Might not be a bad idea to include a .Rproj file with the correct whitespace config in the repo; I assume at least some contributors use RStudio, and I don't think the file would be a problem for those who don't use RStudio.

[MichaelChirico]

Missed the indent change. Might not be a bad idea to include a .Rproj file with the correct whitespace config in the repo; I assume at least some contributors use RStudio, and I don't think the file would be a problem for those who don't use RStudio.

Thanks for the suggestion. Those files are ignored:

data.table/.gitignore

Lines 23 to 24 in 585ec52

	.Rproj.user
	data.table.Rproj

As are several related to other IDEs like Emacs.

Better might be putting in the .dev folder, like our recommended .Rprofile:

https://github.com/Rdatatable/data.table/blob/master/.dev/.Rprofile

There users can get a copy when cloning the repo and just copy it to the package directory.

I am using VScode these days so not sure what an Rproj looks like anymore.

[MichaelChirico]

LGTM, please add a NEWS bullet

[MichaelChirico]

Thanks! Let's merge #6068 first since it's basically ready, then would you mind quickly refactoring this PR to use the new functionality?

[MichaelChirico]

OK, #6068 is merged, would you like to have a go at using the new argument to colnamesInt()?

[sluga]

@MichaelChirico Hmm, the tests I added for this feature pass after the update, but test 1967.63 fails:

Test 1967.63 produced 0 errors but expected 1
Expected: Item 2 of order (1) is either NA, out of range [1,2], or is duplicated. The new order must be a strict permutation of 1:n
Observed:

[MichaelChirico]

Please add a test for setcolorder(DT, c('a', 'b', 'd'), skip_absent=TRUE).

The NEWS item currently reads like 'c' will be dropped from the output, which would be bad -- I would expect the above to be equivalent to setcolorder(DT, c("a", "b")).