Fix spurious printing of forecast/scores objects during :=

[nikosbosse]

CLAUDE:

Summary

• Fixes forecast and scores objects print output every time a column is changed #935 by leveraging data.table's shouldPrint() mechanism to suppress autoprinting when := modifies forecast or scores objects in place
• Adds print.scores() method with the same autoprint suppression
• Skips validation in [.forecast during := operations (the validation was consuming data.table's autoprint suppression flag and triggering the printing side-effect)
• Bumps data.table dependency from >= 1.16.0 to >= 1.17.0 (which contains the upstream fix in Rdatatable/data.table#6631)

See #1161 for detailed root cause analysis.

How it works

The fix has two layers:

1. print.forecast() and print.scores(): Check data.table's internal shouldPrint(x) at the top of the method. When := is used, data.table sets a flag to suppress autoprinting. If the flag indicates suppression, return invisible(x) immediately without printing any output (including the forecast header).

2. [.forecast(): Detect in-place modifications (:=) by checking identical(x, out) && ...length() > 1 and skip validation. Previously, the validation call inside [.forecast would consume the shouldPrint flag before print.forecast could check it.

Breaking change

[.forecast no longer validates during := operations. Users who relied on the warning when := broke a forecast object's contract (e.g., changing predicted to character) will need to call assert_forecast() explicitly. The $<-, [[<-, and [<- assignment methods still validate.

Test plan

• All 765 existing tests pass
• New tests verify := produces no output for all forecast types (binary, quantile, point, sample_continuous, sample_discrete)
• New tests verify := on scores objects produces no output
• New tests verify multiple sequential := operations produce no output
• New tests verify explicit print() still works after :=
• New tests verify x[] force-print still works
• New tests verify [.forecast validates subsets regardless of size (30-row hack removed)
• New tests verify print header and data content are shown for explicit prints
• Existing snapshot tests unchanged

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.99%. Comparing base (4a222da) to head (3b48329).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1163   +/-   ##
=======================================
  Coverage   97.98%   97.99%           
=======================================
  Files          37       37           
  Lines        1984     1992    +8     
=======================================
+ Hits         1944     1952    +8     
  Misses         40       40           

☔ 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.

[nikosbosse]

CLAUDE (and verified by Nikos): Here's a before/after comparison showing the fix in action:

Before (main):

ex <- data.table::copy(example_quantile)
ex[, model := paste(model, "a")]
#> Forecast type: quantile
#> Forecast unit:
#> location, target_end_date, target_type, location_name, forecast_date, model,
#> and horizon
#>
#> Key: <location, target_end_date, target_type>
#>        location target_end_date target_type observed location_name
#>          <char>          <Date>      <char>    <num>        <char>
#>     1:       DE      2021-01-02       Cases   127300       Germany
#>     2:       DE      2021-01-02      Deaths     4534       Germany
#>     3:       DE      2021-01-09       Cases   154922       Germany
#>     4:       DE      2021-01-09      Deaths     6117       Germany
#>     5:       DE      2021-01-16       Cases   110183       Germany
#>    ---
#> 20541:       IT      2021-07-24      Deaths       78         Italy
#> 20542:       IT      2021-07-24      Deaths       78         Italy
#> 20543:       IT      2021-07-24      Deaths       78         Italy
#> 20544:       IT      2021-07-24      Deaths       78         Italy
#> 20545:       IT      2021-07-24      Deaths       78         Italy
#>   ... (full table unexpectedly printed)

The entire 20,545-row forecast object is printed every time a column is modified via :=. This happens for all forecast types and scores objects.

After (fix branch):

ex <- data.table::copy(example_quantile)
ex[, model := paste(model, "a")]
# (no output — matches the behavior of := on a plain data.table)

print(ex)  # explicit print still works as expected
#> Forecast type: quantile
#> Forecast unit:
#> location, target_end_date, target_type, location_name, forecast_date, model,
#> and horizon
#>
#>        location target_end_date target_type observed ...
#>     1:       DE      2021-01-02       Cases   127300 ...
#>    ---
#> 20541:       IT      2021-07-24      Deaths       78 ...

The same fix applies to scores objects and all forecast types (binary, quantile, point, sample_continuous, sample_discrete).

[seabbs]

Let me know when ready for review.

[nikosbosse]

Ready for review!

[seabbs]

Why is not validating in place modification in this PR?

[seabbs]

utils::getFromNamespace("shouldPrint",

this seems like all we need. I don't know why we have the validation turned off for := in this PR?

Does print.scores need to exist if it just next method itself into data.table?

[seabbs]

Skips validation in [.forecast during := operations (the validation was consuming data.table's autoprint suppression flag and triggering the printing side-effect)

I don't think we want to do this?