Regression in displaying of stacktraces with recursive functions. (on ARM?)

[KristofferC]

It seems a lot of the niceness that has been added for "stackoverflow type" of errors (or just generally when a function is called recursively) has been lost:

As an example:

julia> f(x) = g(x)
f (generic function with 1 method)

julia> g(x) = f(x)
g (generic function with 1 method)

1.7:

julia> f(10)
ERROR: StackOverflowError:
Stacktrace:
     [1] f(x::Int64)
       @ Main ./REPL[1]:1
     [2] g(x::Int64)
       @ Main ./REPL[2]:1
--- the last 2 lines are repeated 39990 more times ---
 [79983] f(x::Int64)
       @ Main ./REPL[1]:1

1.8+:

julia> f(10)
ERROR: StackOverflowError:
Stacktrace:
 [1] g(x::Int64)
   @ Main ./REPL[2]:1
 [2] f(x::Int64)
   @ Main ./REPL[1]:1

The

--- the last 2 lines are repeated 39990 more times ---
 [79983] f(x::Int64)

part is lost.

---

Another example:

julia> f(x) = f(x)
f (generic function with 1 method)

1.7:

julia> f(1)
ERROR: StackOverflowError:
Stacktrace:
 [1] f(x::Int64) (repeats 79984 times)
   @ Main ./REPL[1]:1

1.8:

julia> f(1)
ERROR: StackOverflowError:
Stacktrace:
 [1] f(x::Int64) (repeats 2 times)
   @ Main ./REPL[4]:1

1.8 says repeats 2 times which seems wrong..

[Liozou]

I have the correct (full) error message for both cases on 1.8.5, 1.9.3, 1.10 and master so this is probably system specific. My system is

julia> versioninfo()
Julia Version 1.8.5
Commit 17cfb8e65ea (2023-01-08 06:45 UTC)
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 8 × Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-13.0.1 (ORCJIT, skylake)
  Threads: 1 on 8 virtual cores

[inkydragon]

so this is probably system specific

I guess so, too.
Both cases work fine for me on v1.6~v1.10 and master.

> juliaup status
 Default  Channel  Version                       Update
--------------------------------------------------------
          1.6      1.6.7+0.x64.w64.mingw32
          1.7      1.7.3+0.x64.w64.mingw32
          1.8      1.8.5+0.x64.w64.mingw32
          1.9      1.9.4+0.x64.w64.mingw32
          1.10     1.10.0-rc2+0.x64.w64.mingw32

• master: Commit 649982a (1.11.0-DEV.1065)

julia> versioninfo()
Julia Version 1.11.0-DEV.1065
Commit 649982aa09* (2023-12-09 01:05 UTC)
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: 32 × 13th Gen Intel(R) Core(TM) i9-13900HX
  WORD_SIZE: 64
  LLVM: libLLVM-15.0.7 (ORCJIT, goldmont)
  Threads: 1 on 32 virtual cores

[adrhill]

Adding to this, It would be nice if the (repeats $n times) warning was at least emphasized / printed in color:

julia/base/errorshow.jl

Line 765 in ec32170

printstyled(io, " (repeats $n times)"; color=:light_black)

Current state	With color

[fingolfin]

It works fine for me on Linux x86, i.e. it shows the "repeated"

But on macOS with ARM cpu and Linux on a Raspberry (also with ARM CPU) the message is missing.

This is using Julia 1.11.4 in both cases.

[fingolfin]

This is already "broken" (?) in 1.8.5 on ARM macOS. As such I am not sure if it ever actually worked -- i.e. this may be less a regression and more "this feature is not supported on ARM / is only supported on x86" ?

[KristofferC]

But in my post it seemed to work on 1.7

[fingolfin]

But was your post done in a ARM Linux binary? If so then necessarily for Linux as macOS ARM support was introduced in 1.8...

[KristofferC]

Aha, that might be it then.

[fingolfin]

Just tested on the Raspberry Pi (with 64 Linux) and the issue also occurs there.