Add get method, tests for Tuple and AbstractArrays :: Take 2 (#40809)

[Pramodh-G]

As specified in the PR with the same name (#40856 ), I had messed up my fork and had to start with a fresh fork. Sorry for this, I'm trying to improve :) .
I added tests that were mutating, but I couldn't add CartesianIndex(cc: @mcabbott ). I was getting this error while make ing julia. I would be glad to fix this if someone could guide me through it.

error during bootstrap:
LoadError(at "compiler/compiler.jl" line 3: LoadError(at "abstractarray.jl" line 1473: UndefVarError(var=:CartesianIndex)))
jl_undefined_var_error at /home/learnedfool/github_repos/julia/src/rtutils.c:132
jl_eval_global_var at /home/learnedfool/github_repos/julia/src/interpreter.c:141 [inlined]
eval_value at /home/learnedfool/github_repos/julia/src/interpreter.c:187
do_call at /home/learnedfool/github_repos/julia/src/interpreter.c:116
eval_value at /home/learnedfool/github_repos/julia/src/interpreter.c:206
eval_stmt_value at /home/learnedfool/github_repos/julia/src/interpreter.c:157 [inlined]
eval_body at /home/learnedfool/github_repos/julia/src/interpreter.c:575
jl_interpret_toplevel_thunk at /home/learnedfool/github_repos/julia/src/interpreter.c:718
top-level scope at abstractarray.jl:1473
jl_toplevel_eval_flex at /home/learnedfool/github_repos/julia/src/toplevel.c:884
jl_parse_eval_all at /home/learnedfool/github_repos/julia/src/toplevel.c:1010
jl_load_ at /home/learnedfool/github_repos/julia/src/toplevel.c:1057
include at ./boot.jl:367
include at ./compiler/compiler.jl:19
jl_apply at /home/learnedfool/github_repos/julia/src/julia.h:1760 [inlined]
do_call at /home/learnedfool/github_repos/julia/src/interpreter.c:117
eval_value at /home/learnedfool/github_repos/julia/src/interpreter.c:206
eval_stmt_value at /home/learnedfool/github_repos/julia/src/interpreter.c:157 [inlined]
eval_body at /home/learnedfool/github_repos/julia/src/interpreter.c:575
jl_interpret_toplevel_thunk at /home/learnedfool/github_repos/julia/src/interpreter.c:718
top-level scope at compiler/compiler.jl:68
jl_toplevel_eval_flex at /home/learnedfool/github_repos/julia/src/toplevel.c:884
jl_eval_module_expr at /home/learnedfool/github_repos/julia/src/toplevel.c:195 [inlined]
jl_toplevel_eval_flex at /home/learnedfool/github_repos/julia/src/toplevel.c:672
jl_toplevel_eval_in at /home/learnedfool/github_repos/julia/src/toplevel.c:936
eval at ./boot.jl:369
jl_apply at /home/learnedfool/github_repos/julia/src/julia.h:1760 [inlined]
do_call at /home/learnedfool/github_repos/julia/src/interpreter.c:117
eval_value at /home/learnedfool/github_repos/julia/src/interpreter.c:206
eval_stmt_value at /home/learnedfool/github_repos/julia/src/interpreter.c:157 [inlined]
eval_body at /home/learnedfool/github_repos/julia/src/interpreter.c:575
jl_interpret_toplevel_thunk at /home/learnedfool/github_repos/julia/src/interpreter.c:718
top-level scope at compiler/compiler.jl:3
jl_toplevel_eval_flex at /home/learnedfool/github_repos/julia/src/toplevel.c:884
jl_parse_eval_all at /home/learnedfool/github_repos/julia/src/toplevel.c:1010
jl_load_ at /home/learnedfool/github_repos/julia/src/toplevel.c:1057
jl_load at /home/learnedfool/github_repos/julia/src/toplevel.c:1070
exec_program at /home/learnedfool/github_repos/julia/src/jlapi.c:513
true_main at /home/learnedfool/github_repos/julia/src/jlapi.c:565
jl_repl_entrypoint at /home/learnedfool/github_repos/julia/src/jlapi.c:695
main at /home/learnedfool/github_repos/julia/cli/loader_exe.c:51
__libc_start_main at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
_start at /home/learnedfool/github_repos/julia/usr/bin/julia (unknown line)

make[1]: *** [sysimage.mk:61: /home/learnedfool/github_repos/julia/usr/lib/julia/corecompiler.ji] Error 1
make: *** [Makefile:82: julia-sysimg-ji] Error 2

[simeonschaub]

I added tests that were mutating

Maybe I am missing something, but I don't see any tests here where f has side effects.
Supporting CartesianIndex doesn't need to be part of this PR, let's keep that for a future PR.

[Pramodh-G]

Oh, I maybe I misinterpreted what side-effects meant. Could you please provide some examples/cases?

I thought you just wanted confirmation that f gets called only when there is a BoundsError, And hence I added tests that actually changed values in the Matrix.

Since f cannot take in any arguments, it felt hard coming up with testcases where an existing global variable is mutated, since it might mess up other tests; If you want, maybe I could initialize some variables myself?

Supporting CartesianIndex doesn't need to be part of this PR, let's keep that for a future PR.

Sure, I'll remove those comments.

Thanks!

[simeonschaub]

A .+ 1 does not modify A at all and you don't yet that it has been modified. The easiest way to test this is probably doing something like :

c = 0
@test get(() -> (c += 1; 0), ...) == ...
@test c == 1

[Pramodh-G]

Sure, I'll try something like this out.