Could we tag more beginner-task issues?

[pedrorodriguesarantes]

Hi all,

I am helping @igorsteinmacher with his Open Source class at Northern Arizona University (NAU). As part of the course, students contribute to real OSS projects through multiple assignments, and we’re planning to work with data.table this semester.

I noticed there are currently around 13 issues labeled beginner-task. Since we will have several students contributing, it would be very helpful if a few additional suitable issues could be tagged (or possibly suggested).

If there are specific areas that you think are particularly appropriate for first-time contributors (tests, documentation improvements, small bug fixes, or minor enhancement) we would really appreciate the guidance.

Thanks for the attention. We are excited to collaborate and contribute, looking forward to working together!!!!

[MichaelChirico]

Thanks for reaching out! How many students are in this class?

Please note the following important considerations:

• Our review bandwidth is already a bit strained. Low-quality submissions are especially taxing; it would be great if you could incorporate an element of pre-screening on your end (e.g. a peer reviews a PR before submitting it "officially" to the repo)
• We do not accept LLM-generated code: https://github.com/Rdatatable/data.table/blob/master/.github/CONTRIBUTING.md#pull-requests-prs

With those caveats, I employed Gemini to generate the list below, which I've lightly vetted. I don't think it did a very good job but it's directionally acceptable.

Interested contributors are strongly encouraged to first reach out through the issue discussion -- we have some experience of first-time contributors writing PRs for borderline issues that wind up being closed, either because they missed some nuance that's only in the contributors' heads for now, or because they forged ahead on a solution to a stale problem we'd rather just close, etc.

Difficulty 0 - Extremely simple (typos, docs, error message improvements)

• Document Grattan's (Hugh) use of data.table #2862
• Hyperlinks to navigate between Vignettes? #5833
• Revise explanation of 'join-to'/'join-from' in mergelist documentation #7190
• Document side effects of := on keys #7409

Difficulty 1 - Very simple (minor documentation additions, simple refactoring)

• New vignette for data.table munging utilities #4208
• Understanding the value of .I for non-matching rows when using .EACHI #5457

Difficulty 2 - Simple (consistency fixes, minor bug reports, clear clarifications)

• Expression of column names in by: clarify the function of list and parentheses #2391
• Symbol .I consistency when not grouping #2598
• "Can't assign to the same column twice in the same query" message unhelpful, and inconsistent #2967
• Clarification on use of .. (dot dot) #3199
• cutoff old information from FAQ #3275
• vignettes: timeseries - ordered observations #3453
• deprecate multiple columns to be defined as scalar character #4357
• keepLeadingZeros = TRUE can change header behavior #5405
• blank.lines.skip won't work when the top lines is empty #5611
• translated vignette proposal : all pages link back to EN root vignette #6613
• Extend md-lint to ensure vignette() references are links for the website #6638
• Consistently qualify [base] in Rd links #7095
• tests on R 3.4 raises extra warning #7099
• .SD ignores locally updated vectors in j #7431

Difficulty 3 - Straightforward (small features, well-defined bugs, simple R-level enhancements)

• Add use cases of x. prefix in joins to the join vignette #662
• Improve access to data.tables metadata - tables(), columns() #1648
• Reshaping forces to reorder the columns. #2009
• Extend ..var to work in some other cases #2057
• fread for directories #2582
• Keyboard driven pager of data.tables #2893
• Add introductory general principles to benchmarking vignette #3028
• fread: input where every second line is empty gives no warning and only last line is read  #3339
• Warning for set / := when numToDo > nrow(x)? #3557
• dcast's subset argument could behave more like base::subset than reshape2::dcast #4325
• fpaste: fwrite output as a character vector #4572
• question | warnings with joins #4601
• setDT modifies the class of the origin data.frame #4784
• Query on possibilities for fread enhancements for highly structured files #4997
• Don't dispatch [.data.frame if !cedta(); warn/message instead #6337
• Should .SD include a column used to create an ad-hoc 'by'? #6388
• Improve precision/helpfulness of "at an earlier point" error message for selfrefok miss #6494
• Multiple small issues with messages in the C code #6503
• .NATURAL join fails to re-use index #6667
• data.table + sf integration with ggplot2 regression #6707
• Arrow write_parquet removes .internal.selfref, data.table warning message not helpful #6737
• Improve verbose messages in rolling functions #7021
• Use *iB for power-of-2 bytes units consistently #7079
• Add a new helper for setup to minimize boilerplate in a *.Rraw suite #7109
• setnafill error for invalid input should be more helpful #7449
• Could we tag more beginner-task issues? #7649

Difficulty 4 -Moderate-Simple (more involved R-level logic, multiple interactions)

• Consistency of variable names between DT[DT[,list()]] and DT[list()] #827
• copy() on list of data.tables issue #1195
• No realloc warning when adding many inside a function #1633
• Inconsistent and confusing naming convention for unequal join #2006
• New symbol to count rows matched in X[Y] #2124
• Add syntax for "subsetting join" #2158
• Possible inconsistencies in the autonaming and renaming of .N #2165
• dot alias not understood as expected in by/keyby #2447
• Assigning NA_real_ to multiple rows #2959
• variable scoping in grouping when by char vec #3178
• Empty data.table produced with .SD when grouping by all columns #3262
• reconsider treating DT[!int] as DT[-int]? #3699
• := NULL does nothing by group, := .(NULL) fails #3709
• coalesce vectors of type list #3712
• rolling join with a new twist, the mult=unique #3871
• rbindlist edge case inconsistency #4036
• aggregate same variable as groupby on-the-fly expression won't make it length 1 #4079
• String coercion inserts spaces, as.matrix.data.table #4209
• Inconsistent lengths of col when in both by and .SDcols #4317
• Split time-zone specific tests into their own script #4474
• Inconsistent output with j being a data.table and by #4715
• Inconsistent semantics after setDT #4783
• Inconsistent behavior in keyed/unkeyed joins against duplicate columns #4891
• More tests detecting invalid index using selfrefok #5085
• An empty DT() doesn't print at the end of a pipe chain #5106
• CSV header with column name "NA" is incorrectly parsed by fread #5124
• Inconsistency of excluding columns in j vs. .SDcols #5193
• Inconsistent replacement of list column element with NULL in table with 1 row #5558
• i column not found in rolling join with by #5564
• data.table::.N may not work as expected in several places #5958
• More consistency in handling matrices in j #6055
• Don't "double-print" in output= tests #6188
• FR: preserve names of functions inside data.table when by =  #6219
• Test C locale breaks test suite  #6564
• dcast naming: 1 vs. 2+ value.var #6582
• check correct usage of LENGTH/length in C code #6877
• froll.Rraw could use test()'s options verbose #7013
• as.data.frame.data.table(x) errors when x has no column names: setalloccol #7067
• test() failure for missed expected warning not ideal #7092
• as.IDate() drops names #7252

Details

Gemini prompt:

Help fulfil #7649.

Read all issue metadata for the Rdatatable/data.table repo, namely the title and labels.
Ignore issues already labeled with 'beginner-task'.
Filter any issues which look like they obviously require an expert/experienced hand.
Then, for the remaining candidate list of issues, read further details -- the issue body and comments.
Assign a perceived difficulty rating of 0-10 to all issues.
Remove any issue with a perceived difficulty exceeding 4.
Finally, return the list of 0-4-ranked issues, in ascending order.

[pedrorodriguesarantes]

Hi all,

I just want to sincerely thank you for the support and openness the community has shown so far. The list of issues separated by level has being extremely helpful to guide the students. We also appreciate the guidance provided in the issue discussions.

This has been a valuable opportunity for the class, and we are grateful for the collaboration. Thanks again for all the support!!!

[aitap]

Are there any students still in need of a task? Do they need any help? Should we expect 50 PRs the night before the deadline some time in April?

[pedrorodriguesarantes]

@aitap, all the expected pull requests have now been submitted. I would love to hear feedback from the community about the students work with the project.

Should we continue encouraging students to contribute to data.table every semester? Also, is there anything we could improve on our side to make future interactions smoother for the community?

[ben-schwen]

From my perspective, the communication and interaction with the students were above average compared to most first-time OSS contributors. Whether this was mainly due to the additional extrinsic motivation from the course or simply the peer-group dynamic is hard to say.

I think this is ultimately the same trade-off that always exists in open source. If even a few contributors stay involved long term, continue improving, and eventually reduce the review and maintenance burden, then the investment is clearly worth it. If it remains a one time contribution, it becomes more debatable whether the required mentoring and review effort pays off.

[aitap]

The students definitely helped, so many thanks! If there's a neglected PR in need of a review (or otherwise seemingly stuck on our end), please let me know.

Some pull requests need just a few fixes before they can be merged: for #7727 they are purely cosmetic, for #7728 they are straightforward, and for for #7719 they are more substantial. Unfortunately, #7724 did not progress much after the review.