www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Issue with algorithm.fold - dmd hangs

reply Borax Man <rotflol2 hotmail.com> writes:
Sorry if this is an actual bug with DMD.  I suspect it might be, but I'm
not sure whether the fault is DMD or my code.

Short story, I wrote a program for work to do calculations for my team
in D, and use Windows 64bit DMD to compile.  I upgraded to the latest
DMD, and its hangs on compile.  I tried under Linux, and same issue.

Verison is

DMD64 D Compiler v2.111.0
Copyright (C) 1999-2025 by The D Language Foundation, All Rights Reserved
written by Walter Bright

When I compile using dmd with the "-v" switch, these are the last
messages.

import    core.internal.string	(/usr/include/dmd/druntime/import/core/internal/string.d)
semantic3 foldtest
import    std.utf	(/usr/include/dmd/phobos/std/utf.d)
import    std.algorithm.internal	(/usr/include/dmd/phobos/std/algorithm/internal.d)

It hangs at the last line, just stays there consuming CPU cycles.

I've extracted a minimal failing copy.  The line which calls "fold" to
sum the values causes it to hang.  This compiles fine with GDC and LDC2,
and worked with DMD version before 2.100 (I didn't try anything between
2.100 and 2.111.0)

Is my code wrong, or have I stumbled on a bug?

Code below

==================================================

import std.algorithm;
import std.range;
import std.stdio;

final class DataClass {
private:
	int x;

	this()
	{}
	this(in int _x)
	{
		x=_x;
	}
	
	auto ref opOpAssign(string op : "+")(in DataClass other)
	{
		x += other.x;
		return this;
	}

	void printx()
	{
		writeln(x);
	}
	
	auto opBinary(string op : "+")(in DataClass other)
	{
		DataClass newr = new DataClass();
		newr.x = x + other.x;
		return newr;
	}
}

int main()
{
	DataClass[] dataclassarray;
	DataClass result = new DataClass();
	DataClass d1 = new DataClass(3);
	DataClass d2 = new DataClass(4);
	DataClass d3= new DataClass(10);
	DataClass d4 = new DataClass(12);
	DataClass res2 = new DataClass(10);
	dataclassarray~=d1;
	dataclassarray~=d2;
	dataclassarray~=d3;
	dataclassarray~=d4;
	res2 += d1; // This works
	res2.printx();
	res2 = d1 + d2; // This works
	res2.printx();
	result = dataclassarray.fold!((a,b) => a + b);  // This causes dmd to hang.
	result.printx();
	return 0;
}
Apr 04
next sibling parent monkyyy <crazymonkyyy gmail.com> writes:
On Friday, 4 April 2025 at 23:35:14 UTC, Borax Man wrote:

 Sorry if this is an actual bug with DMD.  I suspect it might 
 be, but I'm not sure whether the fault is DMD or my code.

 Short story, I wrote a program for work to do calculations for 
 my team in D, and use Windows 64bit DMD to compile.  I upgraded 
 to the latest DMD, and its hangs on compile.  I tried under 
 Linux, and same issue.

 Verison is

 DMD64 D Compiler v2.111.0
 Copyright (C) 1999-2025 by The D Language Foundation, All 
 Rights Reserved written by Walter Bright

 When I compile using dmd with the "-v" switch, these are the 
 last messages.

 import    
 core.internal.string	(/usr/include/dmd/druntime/import/core/internal/string.d)
 semantic3 foldtest
 import    std.utf	(/usr/include/dmd/phobos/std/utf.d)
 import    
 std.algorithm.internal	(/usr/include/dmd/phobos/std/algorithm/internal.d)

 It hangs at the last line, just stays there consuming CPU 
 cycles.

 I've extracted a minimal failing copy.  The line which calls 
 "fold" to sum the values causes it to hang.  This compiles fine 
 with GDC and LDC2, and worked with DMD version before 2.100 (I 
 didn't try anything between 2.100 and 2.111.0)

 Is my code wrong, or have I stumbled on a bug?

 Code below

 ==================================================

 import std.algorithm;
 import std.range;
 import std.stdio;

 final class DataClass {
 private:
 	int x;

 	this()
 	{}
 	this(in int _x)
 	{
 		x=_x;
 	}

 	auto ref opOpAssign(string op : "+")(in DataClass other)
 	{
 		x += other.x;
 		return this;
 	}

 	void printx()
 	{
 		writeln(x);
 	}

 	auto opBinary(string op : "+")(in DataClass other)
 	{
 		DataClass newr = new DataClass();
 		newr.x = x + other.x;
 		return newr;
 	}
 }

 int main()
 {
 	DataClass[] dataclassarray;
 	DataClass result = new DataClass();
 	DataClass d1 = new DataClass(3);
 	DataClass d2 = new DataClass(4);
 	DataClass d3= new DataClass(10);
 	DataClass d4 = new DataClass(12);
 	DataClass res2 = new DataClass(10);
 	dataclassarray~=d1;
 	dataclassarray~=d2;
 	dataclassarray~=d3;
 	dataclassarray~=d4;
 	res2 += d1; // This works
 	res2.printx();
 	res2 = d1 + d2; // This works
 	res2.printx();
 	result = dataclassarray.fold!((a,b) => a + b);  // This causes 
 dmd to hang.
 	result.printx();
 	return 0;
 }


works for me
Apr 04
prev sibling next sibling parent reply "Richard (Rikki) Andrew Cattermole" <richard cattermole.co.nz> writes:
Works with dmd 2.109.1 for me.

I use MSVC to link.

Do you have Visual studio installed, or are you using the bundled LLD?
Apr 04
parent reply Borax Man <rotflol2 hotmail.com> writes:
On 2025-04-05, Richard (Rikki) Andrew Cattermole <richard cattermole.co.nz>
wrote:

 Works with dmd 2.109.1 for me.

 I use MSVC to link.

 Do you have Visual studio installed, or are you using the bundled LLD?


Only using the bundled LLD.  For what its worth, I used the .7z archive
and extracted it on the Windows machine.
Apr 04
parent Borax Man <rotflol2 hotmail.com> writes:
On 2025-04-05, Borax Man <rotflol2 hotmail.com> wrote:

 On 2025-04-05, Richard (Rikki) Andrew Cattermole <richard cattermole.co.nz>
wrote:

 Works with dmd 2.109.1 for me.

 I use MSVC to link.

 Do you have Visual studio installed, or are you using the bundled LLD?


 Only using the bundled LLD.  For what its worth, I used the .7z archive
 and extracted it on the Windows machine.


I tried with DMD 2.110.0 for Linux, and it compiled and ran fine.  I now
suspect a bug in dmd 2.111.0.  Note that when I replace "fold" with
"reduce", the compile still hangs.

I don't have a github account to file a bug report.
Apr 06
prev sibling parent Paul Backus <snarwin gmail.com> writes:
On Friday, 4 April 2025 at 23:35:14 UTC, Borax Man wrote:

 Sorry if this is an actual bug with DMD.  I suspect it might 
 be, but I'm not sure whether the fault is DMD or my code.

 Short story, I wrote a program for work to do calculations for 
 my team in D, and use Windows 64bit DMD to compile.  I upgraded 
 to the latest DMD, and its hangs on compile.  I tried under 
 Linux, and same issue.


I can reproduce this on 64-bit Linux with DMD 2.111.0.
Apr 04