anyhow::bail!("Maybe we should never reach this?"): what if we did reach this?

[mortenpi]

I have an MWE that reaches this line:

juliaup/src/bin/julialauncher.rs

Line 327 in 33d1fdb

anyhow::bail!("Maybe we should never reach this?");

It involves interrupting an IO hang due to --warn-overwrite=yes. This came up in JuliaDocs/Documenter.jl#2121. To reproduce, run the following script with IOCapture 0.2.2 (0.2.3 contains JuliaDocs/IOCapture.jl#15 which removes the hang):

# julia --warn-overwrite=yes --project script.jl
using IOCapture
println("The following will hang forever... Ctrl+C now.")
c = IOCapture.capture() do
    for i in 1:1024
        eval(:(function f() 1 end))
    end
end

This must be run as a script, i.e. julia --warn-overwrite=yes --project script.jl (where julia is juliaup). I have the correct manifest here: https://gist.github.com/mortenpi/6223ef6f3c02a1b9fbfe2530a0d884e3

It should result in

$ julia --warn-overwrite=yes --project script.jl
The following will hang forever... Ctrl+C now.
^CError: Maybe we should never reach this?

This is on Linux. And it seems like the Julia version matters -- I get this with the official 1.9 binary, but not with 1.7 or 1.8. I didn't try 1.10/master.

julia> versioninfo()
Julia Version 1.9.0
Commit 8e630552924 (2023-05-07 11:25 UTC)
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 16 × 12th Gen Intel(R) Core(TM) i7-1260P
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-14.0.6 (ORCJIT, alderlake)
  Threads: 1 on 16 virtual cores

[davidanthoff]

@maleadt and @staticfloat any thoughts? I think you helped me (i.e. told me ;) how to implement the signal handling stuff on Linux, and this seems like a corner case we aren't handling correctly yet?

[staticfloat]

Okay so it looks like what's happening here is that Julia is dying with a signal (SIGINT) and when rust tries to kill itself with that same signal it doesn't die. I believe that's because of the ctrlc handler set here. We should revert our SIGINT handler back to the default behavior before calling the raise() calls.

I wish ctrlc contained a revert_to_default(), but it doesn't look like it does: Detegr/rust-ctrlc#106

[LilithHafner]

I ran into this when ctrlc ing a package compiler run

Details

x@x julia % julia --project compile.jl 
PackageCompiler: bundled libraries:
  ├── Base:
  │    ├── libLLVM.dylib - 60.192 MiB
  │    ├── libatomic.1.dylib - 108.375 KiB
  │    ├── libdSFMT.dylib - 85.438 KiB
  │    ├── libgcc_s.1.1.dylib - 169.828 KiB
  │    ├── libgfortran.5.dylib - 1.788 MiB
  │    ├── libgmp.10.dylib - 474.734 KiB
  │    ├── libgmpxx.4.dylib - 112.391 KiB
  │    ├── libgomp.1.dylib - 420.859 KiB
  │    ├── libjulia-codegen.1.9.dylib - 1.711 MiB
  │    ├── libjulia-internal.1.9.dylib - 4.132 MiB
  │    ├── libmpfr.6.dylib - 543.938 KiB
  │    ├── libopenlibm.4.0.dylib - 174.516 KiB
  │    ├── libpcre2-8.0.dylib - 758.344 KiB
  │    ├── libquadmath.0.dylib - 373.531 KiB
  │    ├── libssp.0.dylib - 87.641 KiB
  │    ├── libstdc++.6.dylib - 3.417 MiB
  │    ├── libuv.2.dylib - 359.312 KiB
  │    ├── libz.1.2.13.dylib - 171.125 KiB
  │    ├── libunwind.1.0.dylib - 108.562 KiB
  │    ├── libjulia.1.9.dylib - 222.281 KiB
  ├── Stdlibs:
  │   ├── LibCURL_jll
  │   │   ├── libcurl.4.dylib - 581.734 KiB
  │   ├── LibGit2_jll
  │   │   ├── libgit2.1.5.0.dylib - 1.199 MiB
  │   ├── LibSSH2_jll
  │   │   ├── libssh2.1.0.1.dylib - 273.797 KiB
  │   ├── MbedTLS_jll
  │   │   ├── libmbedcrypto.2.28.2.dylib - 608.562 KiB
  │   │   ├── libmbedtls.2.28.2.dylib - 308.484 KiB
  │   │   ├── libmbedx509.2.28.2.dylib - 202.234 KiB
  │   ├── OpenBLAS_jll
  │   │   ├── libopenblas64_.0.3.21.dylib - 21.465 MiB
  │   │   ├── libopenblas64_.dylib - 21.465 MiB
  │   ├── SuiteSparse_jll
  │   │   ├── libamd.2.4.6.dylib - 100.938 KiB
  │   │   ├── libbtf.1.2.6.dylib - 83.828 KiB
  │   │   ├── libcamd.2.4.6.dylib - 100.969 KiB
  │   │   ├── libccolamd.2.9.6.dylib - 117.328 KiB
  │   │   ├── libcholmod.3.0.14.dylib - 978.078 KiB
  │   │   ├── libcholmod.dylib - 978.078 KiB
  │   │   ├── libcolamd.2.9.6.dylib - 100.672 KiB
  │   │   ├── libklu.1.3.8.dylib - 250.406 KiB
  │   │   ├── libldl.2.2.6.dylib - 84.344 KiB
  │   │   ├── librbio.2.2.6.dylib - 117.922 KiB
  │   │   ├── libspqr.2.0.9.dylib - 289.922 KiB
  │   │   ├── libspqr.dylib - 289.922 KiB
  │   │   ├── libsuitesparseconfig.5.10.1.dylib - 84.312 KiB
  │   │   ├── libumfpack.5.7.9.dylib - 799.469 KiB
  │   │   ├── libumfpack.dylib - 799.469 KiB
  │   ├── libblastrampoline_jll
  │   │   ├── libblastrampoline.5.dylib - 2.297 MiB
  │   ├── nghttp2_jll
  │   │   ├── libnghttp2.14.dylib - 285.500 KiB
  Total library file size: 129.002 MiB
⠙ [00m:00s] PackageCompiler: compiling base system image (incremental=false)^C
[58518] signal (2): Interrupt: 2
in expression starting at /Users/x/.julia/dev/related_post_gen/julia/compile.jl:3
kevent at /usr/lib/system/libsystem_kernel.dylib (unknown line)
unknown function (ip: 0x0)
Allocations: 5491585 (Pool: 5487484; Big: 4101); GC: 8
ERROR: rethrow() not allowed outside a catch block
^[[A┌ Warning: temp cleanup
│   exception =
│    schedule: Task not runnable
│    Stacktrace:
│      [1] error(s::String)
│        @ Base ./error.jl:35
│      [2] enq_work(t::Task)
│        @ Base ./task.jl:769
│      [3] yield
│        @ ./task.jl:882 [inlined]
│      [4] yield
│        @ ./task.jl:880 [inlined]
│      [5] Channel{Tuple{String, Vector{String}, Vector{String}}}(func::Base.Filesystem.var"#28#31"{String}, size::Int64; taskref::Nothing, spawn::Bool)
│        @ Base ./channels.jl:140
│      [6] Channel (repeats 2 times)
│        @ ./channels.jl:132 [inlined]
│      [7] #walkdir#27
│        @ ./file.jl:967 [inlined]
│      [8] walkdir
│        @ ./file.jl:928 [inlined]
│      [9] prepare_for_deletion(path::String)
│        @ Base.Filesystem ./file.jl:507
│     [10] temp_cleanup_purge(; force::Bool)
│        @ Base.Filesystem ./file.jl:542
│     [11] temp_cleanup_purge
│        @ ./file.jl:535 [inlined]
│     [12] (::Base.var"#930#931")()
│        @ Base ./initdefs.jl:354
│     [13] _atexit(exitcode::Int32)
│        @ Base ./initdefs.jl:416
└ @ Base.Filesystem file.jl:547
schedule: Task not runnable
Stacktrace:
^CError: Maybe we should never reach this?