graphemes(s, m:n) substring slicing

[stevengj]

People seem to want this functionality, and it's nontrivial to implement efficiently & correctly.

(Discoverability is a challenge here, because most people wanting to access a "slice of characters" in Unicode probably don't know what a grapheme is, but at least this gives us a succinct answer to such questions.)

[KristofferC]

Should anything be said about the time complexity of this operation in the docstring?

[stevengj]

Should anything be said about the time complexity of this operation in the docstring?

Sure, added.

[stevengj]

buildkite CI errors seem unrelated:

LoadError("sysimg.jl", 19, LoadError("/cache/build/dockerized-amdci4-8/julialang/julia-master/tmp/test-asan/asan/usr/share/julia/stdlib/v1.9/Random/src/Random.jl", 3, LoadError("/cache/build/dockerized-amdci4-8/julialang/julia-master/tmp/test-asan/asan/usr/share/julia/stdlib/v1.9/Random/src/DSFMT.jl", 3, InexactError(:trunc, Int64, Inf))))

and

LibGit2/libgit2                          (13) \|         failed at 2022-02-19T22:52:54.464
--
  | Error During Test at /cache/build/default-amdci5-8/julialang/julia-master/julia-db411f4927/share/julia/stdlib/v1.9/LibGit2/test/libgit2.jl:2224
  | Got exception outside of a @test
  | IOError: read: i/o error (EIO)

and

SuiteSparse                              (9) \|  1907.82 \|  11.17 \|  0.6 \|    9577.14 \|  2255.55
--
  | Profile                                  (5) \|         failed at 2022-02-20T00:35:48.352
  | Test Failed at /cache/build/default-amdci4-6/julialang/julia-master/julia-db411f4927/share/julia/stdlib/v1.9/Profile/test/runtests.jl:165
  | Expression: getline(values(fdictc)) == getline(values(fdict0)) + 2
  | Evaluated: nothing == 24
  | Test Failed at /cache/build/default-amdci4-6/julialang/julia-master/julia-db411f4927/share/julia/stdlib/v1.9/Profile/test/runtests.jl:187
  | Expression: parse(Int, s) > 100
  | Evaluated: 68 > 100

[matthieugomez]

Isn't it a bit confusing that graphemes(x, m:n) returns a SubString but graphemes(x) does not? I guess it'd be better to have a different name.