benchmark-node-micro-benchmarks is broken

[panva]

see e.g. https://ci.nodejs.org/view/Node.js%20benchmark/job/benchmark-node-micro-benchmarks/1872

...
...
...

++ cat output160626-153457.csv
++ Rscript benchmark/compare.R
Error: package or namespace load failed for ‘ggplot2’ in dyn.load(file, DLLpath = DLLpath, ...):
 unable to load shared object '/usr/local/lib/R/site-library/rlang/libs/rlang.so':
  /usr/local/lib/R/site-library/rlang/libs/rlang.so: undefined symbol: SETLENGTH
Execution halted

[sxa]

Has something changed in the requirements of the benchmarks? There's not obvious indication that the referenced .so library has been updated on the machine recently.

[richardlau]

Has something changed in the requirements of the benchmarks? There's not obvious indication that the referenced .so library has been updated on the machine recently.

R and the R modules (ggplot2, plyr) are installed by our Ansible playbook, and the most likely explanation for change is that things got updated when the playbook was run recently with changes to update the machines to run on Java 25.

Given the error message, it is likely to be related to this change in R 4.6.0:

https://cran.r-project.org/doc/manuals/r-release/NEWS.html

• These entry points are now marked as hidden and no longer have declarations in installed header files: ENVFLAGS, EXTPTR_PROT, FRAME, ENCLOS, EXTPTR_PTR, EXTPTR_TAG, HASHTAB, IS_S4_OBJECT, LEVELS, NAMED, OBJECT, R_shallow_duplicate_attr, Rf_isValidString, Rf_lazy_duplicate, Rf_NonNullStringMatch, SETLENGTH, SETLEVELS, SET_BODY, SET_CLOENV, SET_ENCLOS, SET_FORMALS, SET_ENVFLAGS, SET_FRAME, SET_GROWABLE_BIT, SET_HASHTAB, SET_NAMED, SET_S4_OBJECT, SET_TRUELENGTH, STDVEC_DATAPTR, TRUELENGTH, UNSET_S4_OBJECT, XTRUELENGTH.

[aduh95]

FWIW you can still get the benchmark results by pasting the CI CSV output to the stdin of nix-shell -I nixpkgs=./tools/nix/pkgs.nix --pure -E '(import <nixpkgs> {}).mkShell { buildInputs = import ./tools/nix/benchmarkTools.nix { withHttpBenchmarkDeps = false; }; }' --keep LC_ALL --keep LANG --run 'Rscript benchmark/compare.R'