⚠️ GitHub issue #43044 has been automatically assigned in GitHub to PR creator.

I think this is ok, just one suggestion for something else to remove. AFAICT we only call r_string_size on a STRING_ELT from a STRSXP, and similarly for utf8_string/utf8_strings. Rf_translateCharUTF8 should always work on those inputs.

Here's the R internals version of IS_ASCII etc.; LEVELS is defined as sxpinfo.gp, an opaque bitmask. Even if this change turns out not to be correct, I don't think we want to be touching sxpinfo.gp for anything ourselves. https://github.com/wch/r-source/blob/0ebd0b0b574847597ac19fbf0bbbf31e3197e84b/src/include/Defn.h#L832-L853

Ah great, you've found the line in the scriptures to confirm that #43044 (comment)

Ah great, you've found the line in the scriptures to confirm that #43044 (comment)

It took many nights of studying the holy text

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit ca49843.

There were 5 benchmark results indicating a performance regression:

• Commit Run on ec2-m5-4xlarge-us-east-2 at 2024-07-07 18:35:04Z
  • dataframe-to-table (R) with dataset=chi_traffic_2020_Q1, language=R
  • file-write (R) with compression=uncompressed, dataset=fanniemae_2016Q4, file_type=feather, input_type=dataframe, language=R
• and 3 more (see the report linked below)

The full Conbench report has more details. It also includes information about 11 possible false positives for unstable benchmarks that are known to sometimes produce them.