`stop_native_implementation()`/`vec_ptype2.factor.factor()` error

[pbbgss]

Error

Error in `stop_native_implementation()`:
  ! `vec_ptype2.factor.factor()` is implemented at C level.
This R function is purely indicative and should never be called.
ℹ This is an internal error that was detected in the vctrs package.
Please report it at <https://github.com/r-lib/vctrs/issues> with a reprex and the full backtrace.
Run `rlang::last_trace()` to see where the error occurred.

Backtrace

Backtrace:
    ▆
 1. └─dplyr::bind_rows(iris_1, iris_1)
 2.   └─vctrs::vec_rbind(!!!dots, .names_to = .id, .error_call = current_env())
 3.     └─vctrs (local) `<fn>`()
 4.       └─vctrs:::vec_ptype2.factor.factor(...)
 5.         └─vctrs:::stop_native_implementation("vec_ptype2.factor.factor")
 6.           └─cli::cli_abort(...)
 7.             └─rlang::abort(...)

REPREX

install.packages("renv")
renv::install("dplyr@1.1.4")
renv::install("vctrs@0.6.5")

# Produces the error
iris_1 <- iris 
attr(iris_1$Species, "class") <- c("factor", "foo_factor")
dplyr::bind_rows(iris_1, iris_1)

# Doesn't produce the error
attr(iris_1$Species, "class") <- c("foo_factor", "factor")
dplyr::bind_rows(iris_1, iris_1)

[brookslogan]

The likely pathway is

• various delegators
  • vec_ptype2_impl
    • either arg S3
      • vec_ptype2_dispatch_native
        • not bare factors
        • signal inapplicability of native dispatch
      • Continue on since native dispatch failed
    • Launch second pass on prototypes
      • Second pass encounters same issues, falls back on S3

This could potentially be resolved by either:

• Having vec_ptype2_dispatch_native replace checks regarding bare factors with checks on whether the first entry of the class vector is "factor", since the vctrs dispatch model appears to ignore the later entries of the class vector in standard operation. (Though was it intended to be S3-equivalent? That seems like it would be helpful in some cases.)
• Having vec_ptype2.factor.factor strip the classes down to bare factors and re-delegate to the impl.