fix auto-print detection on R 3.4/3.5 where sys.calls() returns a promise

[skitsy24]

Closes #7099.

On R 3.4 and 3.5, the top-level call captured via sys.calls() during auto-print is not a direct reference to base::print, but a promise object that evaluates to it. As a result, the existing check:

identical(SYS[[1L]][[1L]], print)

fails, and print.data.table() incorrectly treats auto-print as an explicit print() call. This leads to unintended output in cases where printing should be suppressed (e.g., after := assignments).

Thanks to @aitap for identifying the root cause.

My fix adds a check on whether print() is wrapped in a "promise" to differentiate between the older R versions. If it is a promise, the promise is evaluated, and then checked against print.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.04%. Comparing base (3a45206) to head (194ef2b).
⚠️ Report is 5 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #7727   +/-   ##
=======================================
  Coverage   99.04%   99.04%           
=======================================
  Files          87       87           
  Lines       17123    17043   -80     
=======================================
- Hits        16959    16880   -79     
+ Misses        164      163    -1     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[aitap]

This disappeared in R-3.6.0, right? Let's wrap this branch in # nocov start ... # nocov end and add a # TODO(R>=3.6) comment to remove it once we raise the minimal supported R version.

[aitap]

The fix is good, tested manually on R-3.5.0. Could you please fix the whitespace issues and add a NEWS.md item describing the bug fix?

[skitsy24]

Changes made, and added item to NEWS

[ben-schwen]

pls dont commit Rcheck files

[ben-schwen]

The diff somehow got blown off. Seems like a mislucked merge

[aitap]

You seem to have ended up with master rebased on top of your PR branch. That's the opposite of how rebase is typically used. How did that happen? Did you merge master and then rebase your branch? To recover, try

git checkout fix7099
git remote add upstream https://github.com/rdatatable/data.table
git fetch upstream
git rebase -i upstream/master
# drop 9d7b4f4f Revert workflow file change
git push -f

Normally we only merge master into the PR branches because this is less disruptive than history rewriting: every time you rebase, other people have to really carefully reset --hard their copies of your branch, reapply their changes on top, and hope nothing gets lost in the process. rebase shines in projects like the Linux kernel where every patch resubmission comes from effectively a different branch.

[skitsy24]

Looking at it, I did merge master then rebased my branch, sorry! It should now ONLY have the changes I made, again sorry!