Hello,

Today I came across a strange bug while using D with dmd. I have still not been able to figure out under what conditions does it happen but it seems to be a DMD related bug to me. Here is a reproducible snippet of the code

import std;

alias DG = void delegate();

class TType
{
}

class MyClass
{
    this(TType t1, TType, double, double[2], double, double, DG, TType, TType,
            DG, DG, DG, double, double, double, double, double, ulong, bool)
    {
        assert(t1 is null); // I am passing null so should be null!
        // NOTE: Seems to work in LDC but fails in DMD.
        writeln("No Bug!");
    }
}

void main()
{
    auto tt = new TType;

    new MyClass(null, tt, 0, [0, 0], 0, 0, null, null, null, null, null, null,
            0, 0, 0, 0, 0, 0, false);
}

The code gives an assertion failure on the current versions of dmd (reproducible on run.dlang.io as well) and does not happen when using LDC. The bug seems to be sensitive to the number of arguments and their types making it reproducible only in very limited cases. I have tried my best to reduce it to minimum but still does require these many arguments. The end results seems to me like variables are shifted i.e. variable 1 gets value of variable 2 and so on, but don't have enough proof to support this.

I just wanted some help on the best way to avoid this bug in my code and maybe some clue on what causes the error in the first place and how should I go about reporting this.

Keivan

On Tuesday, 1 November 2022 at 15:40:04 UTC, Keivan Shah wrote:

Could be there's some restriction in DMD on number of arguments.

May I ask if this was just an experiment? I hope you're not having code like that in the wild 🙏

On Tuesday, 1 November 2022 at 15:42:43 UTC, Imperatorn wrote:

Possible, but I think I have had code with more arguments than this and it has worked 😅

Unfortunately, not an experiment. Although have replaced the types so seems silly now, this is part of my constructor for a huge co-coordinator class that takes too many configurable start time parameters and so need to pass these many arguments.

Keivan

On Tuesday, 1 November 2022 at 15:49:54 UTC, Keivan Shah wrote:

Hehe.

One simple thing you could do is to create a struct instead for you params and pass that

On Tuesday, 1 November 2022 at 16:06:44 UTC, Imperatorn wrote:

Yeah, can do, thanks for the suggestion. But anyways still want to see if anyone else has seen this issue, or has a clue about what could be happening here. Seems like a rare issue and took up too much of my time so better if it's solved or at least documented somewhere in the meantime. I have filed an issue for this now: https://issues.dlang.org/show_bug.cgi?id=23450

On Tuesday, 1 November 2022 at 15:49:54 UTC, Keivan Shah wrote:

Regarding the too many configurable parameters, the general advice is to group all of the params into a struct and pass that instead as argument to constructor

Pretty wild bug though, definitely a backend thing

On 11/1/22 11:40 AM, Keivan Shah wrote:

import std;

alias DG = void delegate();

class TType
{
}

class MyClass
{
     this(TType t1, TType, double, double[2], double, double, DG, TType, TType,
             DG, DG, DG, double, double, double, double, double, ulong, bool)
     {
         assert(t1 is null); // I am passing null so should be null!
         // NOTE: Seems to work in LDC but fails in DMD.
         writeln("No Bug!");
     }
}

void main()
{
     auto tt = new TType;

     new MyClass(null, tt, 0, [0, 0], 0, 0, null, null, null, null, null, null,
             0, 0, 0, 0, 0, 0, false);
}

100% this is a bug in DMD. It should be filed.

I ran some more tests, removing almost any of the parameters or changing their types seems to avoid the problem.

I also added a parameter name for the second parameter, and DMD appears to be in this case passing the parameters in the wrong order (t1 is actually tt, t2 is null)

You can also remove the import std, just the assert is enough.

Please file if you can: https://issues.dlang.org

-Steve

This reminds me of an issue i reported last year...

https://issues.dlang.org/show_bug.cgi?id=22583

On Tuesday, 1 November 2022 at 16:39:57 UTC, Steven Schveighoffer wrote:

Hey, have already filled an issue: #23450

On Tuesday, 1 November 2022 at 17:05:03 UTC, ryuukk_ wrote:

This seems to be very similar to the bug I am facing, mostly the same underlying issue. Should we somehow link the 2 issues and escalate?

Link to my issue: #23450