While i like Zig, it has many ergonomics issues that i just can't deal with.. and it has the tendency to make your code unnecessary verbose

- no operator overloading for your math type, you end up chaining methods..

- no function overloading, you have to do things like 'add(comptime T: type, x: antype, y: anytype)`

then: add(f32, myX, myY);

then be prepared with a shit ton of @floatCast(), it can't guess what type you use so you forced to be doubly explicit..

you end up with code that is barely readable

- unused as error, this one hurts a lot.. you can say goodbye to fast iteration time, if you make a game it'll be painful

Zig is still great, i plan to use it for some project, knowing how to use more language helps train your skills and understand pros/cons of features better

You can crosscompile with D too

    ldc2 -mtriple=x86_64-windows-msvc -c foo.d

You can link without the runtime if size is a concern, i haven't had any issues with it

- float default is NaN, why... just WHY!!!!!

- char default is 0xFF....

when you expect things to be just 0, it is a pain to deal with and to remember...

Also the standard library, while useful, i wish it would make use of allocators

They have ``std.experimental.allocator`` packages, for some reason it still is experimental..

There is no tagged union.. you have to import a package for it ``std.sumtype``.. wich is bad when all other languages have built in support for that

Walter if you read this, please! union are useful, but lack of tagged union mean potential bugs in your union when you pick the wrong value...

Proposal for you:

  enum MyTag { A, B, C}

  union MyTaggedUnion: MyTag
  {
      DataA A,
      DataB B,
      DataC C
  }

  struct DataA{}
  struct DataB{}
  struct DataC{}

  auto tgu = MyTaggedUnion.B;
  switch (tgu)
  {
   case A:
   tgu. /* implicit DataA */
   break;
   // need to implement every tags, or error
  }

> char default is 0xFF....

Invalid values, perhaps? Other than making initialization mandatory, this seems like a reasonable way of catching them. For integer types it's not really possible because all their values are valid.

  import std.sumtype;
  
  void main(){
   struct A { int a; }
   struct B { string b; }
   struct C { bool c; }
   
   alias TaggedUnion = SumType!(A, B, C);
   
   auto tgu = TaggedUnion(B("hi"));
   int i = tgu.match!(
    (A a) => a.a,
    (B b) => b.b.length,
    (C c) => c.c * 5
   );
   assert(i == 2);
  }

I know of sumtype, and i think it is a mistake for D to rely on this for such important language feature

I'm not a language dev, so i can't do much to help, using switch/union/enum is more natural, is cleaner, and is easier to add support for IDEs, everyone on the same page

That's in areas like this where the language falls short, and i can see people instead choosing alternatives

https://github.com/dlang/phobos/blob/master/std/sumtype.d

a mess of templates and imports to std.traits

using this will make your compile speed tank.. another reason to avoid

it's very sad that people recommend this instead of asking for built-in version, makes me sometimes reconsider my choice to use the language

if they expect language improvements to be templates, what atrocity will be added next?

this is on the same level as std::bool from C++

> The good news is, I have not put a lot of effort so far into micro-optimizing the compile-time performance of match, so there is almost certainly room for improvement.

it is shame that it ended up in the standard library

once Walter will be gone, i will have no faith in the language anymore

All operations that are fed NaN as an operand produce a NaN result. This makes it very obvious when the initialization of a float has been neglected. Defaulting to 0.0 means uninitialization bugs are nearly impossible to detect.

> char default is 0xFF....

Same thing. It's intended to flush out uninitialization bugs.

Disclaimer, I've never used Zig, so the following are not comparative.

> Is it able to produce as efficient and small binaries,

There are compiler flags to disable linking to the D standard library and D runtime if you need small binaries (you can still use templates that end up in you binary) though most of the time that is more of a loss than gain (in terms of productivity), its not bloated unless you abuse templates a lot as would happen in C++. As for speed of execution, don't use DMD if you care about the it. LDC (the LLVM D Compiler) and GDC (which is part of the GCC) are both strong optimisers.

cross-compile to most platforms,

Yes, use LDC. GDC, like GCC, is not a cross compiler by default though you can build it to be, DMD is X86(64) only.

> metaprogramming etc?

Oh, yes. Compile time function evaluation (referred to usually as CTFE) in combination with mixins (interpret string as code), `static foreach`, `static if` and (much safer than C++) templates makes for a very potent package.

How is it a "scripting language" if you don't want a GC? I use D for the vast majority of my scripting these days, but it would be odd for me to avoid the GC. I would not even think about using it if I had to deal with that.

You can certainly use the std without relying on GC in languages like Rust, Swift, etc, but the important thing is that you can’t in D. I’m sure it can be technically done, but nobody has actually put the effort to do it.

Once finished, everything in the std will be able to make use of it

If you can't wait, you can use this package already with the allocators: https://github.com/dlang-community/containers

Note that many game engines do use garbage collection for this purpose

GC only exist if you call ``new MyThing``

Just use your allocators with malloc/free, that is what i do (you do the same in zig, allocators)

if you want to enforce no GC use in your APIs, there is the @nogc attribute you can use

If you do that, is it RAII like C++/Rust? Objects get destructed at end of scope, and members get destructed recursively? Or do you have to manually call `free()` on everything? (which, granted, is significantly easier in languages with `defer` mechanisms)

    scope(exit):
    scope(failure):
    scope(success):

You can do RAII with D too, it's your choice, based on your use case

We really need to get actual reference counting into the language to cover classes properly as well.

It has been on my todo list to write a DIP for this for a while now. ~rikki.

  scope obj = new Object;
  ...
  // destroyed at end of scope