Incorrect recursive expression crashes the compiler

[Bruce Perens]

I have an incorrect recursive expression that crashes the compiler.
Compiler version:

Crystal 1.12.1 [4cea10199] (2024-04-11)

LLVM: 15.0.7
Default target: x86_64-unknown-linux-gnu

Code:

struct Point
  x : Float64

  def x=(v)
    self.x = v # This is incorrect recursion.
  end

  def initialize(@x = 0.0)
  end
end

a = Array(Point).new(1)

p = (a.to_unsafe + 0).value.x = 

Result:

bruce@bruce16:~$ crystal foo.cr
Stack overflow (e.g., infinite or very deep recursion)
[0x55f6003b1bf6] *Exception::CallStack::print_backtrace:Nil +118 in /home/bruce/.cache/crystal/crystal-run-foo.tmp
[0x55f6003a06e6] ~procProc(Int32, Pointer(LibC::SiginfoT), Pointer(Void), Nil) +310 in /home/bruce/.cache/crystal/crystal-run-foo.tmp
[0x7f8a7f784510] ?? [phone] in /lib/x86_64-linux-gnu/libc.so.6
[0x55f600453dd1] *Point#x=<Int32>:NoReturn +1 in /home/bruce/.cache/crystal/crystal-run-foo.tmp
[0x55f600453dd6] ?? [phone] in /home/bruce/.cache/crystal/crystal-run-foo.tmp (523510 times)
[0x55f600390c8a] ?? [phone] in /home/bruce/.cache/crystal/crystal-run-foo.tmp
[0x55f600454d06] *Crystal::main_user_code<Int32, Pointer(Pointer(UInt8))>:Nil +6 in /home/bruce/.cache/crystal/crystal-run-foo.tmp
[0x55f600454c7a] *Crystal::main<Int32, Pointer(Pointer(UInt8))>:Int32 +58 in /home/bruce/.cache/crystal/crystal-run-foo.tmp
[0x55f60039e1a6] main +6 in /home/bruce/.cache/crystal/crystal-run-foo.tmp
[0x7f8a7f76f6ca] ?? [phone] in /lib/x86_64-linux-gnu/libc.so.6
[0x7f8a7f76f785] __libc_start_main +133 in /lib/x86_64-linux-gnu/libc.so.6
[0x55f600390771] _start +33 in /home/bruce/.cache/crystal/crystal-run-foo.tmp
[0x0] ???

[George Dietrich]

What is your expected result? To me it seems the crash is somewhat legitimate given the setter for x is calling the setter for x. Maybe you intended to do this instead?

def x=(v)
  @x = v
end

EDIT: Or even shorter:

def x=(@x); end

Or just:

setter x

[Bruce Perens]

Yes. I am just assuming the compiler should never crash no matter what stupid thing I do. At Pixar, a technical director had to know 19 computer languages. After I trained as one, I went back to fixing their software. I have now programmed in so many that my 66-year-old brain confuses them, especially if I've been using another one recently.

…

On Tue, Apr 23, 2024 at 11:00 AM George Dietrich ***@***.***> wrote: What is your expected result? To me it seems the crash is somewhat legitimate given the setter for x is calling the setter for x. Maybe you intended to do this instead? def x=(v) @x = vend — Reply to this email directly, view it on GitHub <#14531 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAINM3YM2DM6TQABOSRXQGTY62OTRAVCNFSM6AAAAABGVMIIX2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANZTGA2TCNZYGQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- Bruce Perens K6BP

[Johannes Müller]

I am just assuming the compiler should never crash no matter what
stupid thing I do.

Indeed, it should not.

But it's not the compiler crashing here. It's your program which runs out of memory due to stack exhaustion.
You can verify by separating building and running the exectuable. The compiler build step should complete just fine.

[Bruce Perens]

Oops. OK, closing the report.