• Cecil Ward (5/5) Jun 25 2023 This is not supported in D and to concatenate string literals an
• Richard (Rikki) Andrew Cattermole (2/2) Jun 25 2023 This use to work.
• Cecil Ward (7/9) Jun 25 2023 I think so too. I recently had to rework a lot of old code
• Dave P. (10/20) Jun 25 2023 It’s easy to make this mistake:
• Cecil Ward (2/23) Jun 25 2023 An excellent point. I now fall in with the new status quo. :-)
• Mathias LANG (8/11) Jun 25 2023 And it's exactly the reasoning behind the deprecation.

Cecil Ward <cecil cecilward.com> writes:

This is not supported in D and to concatenate string literals an 
explicit ‘~’ operator is required. I’m in two minds as to whether 
that is a good thing, and am not settling either way. I would be 
interested to know what the arguments were for and against, when 
this design aspect was chosen.

Richard (Rikki) Andrew Cattermole <rikki cattermole.co.nz> writes:

This use to work.

It was removed a few years ago.

Cecil Ward <cecil cecilward.com> writes:

On Sunday, 25 June 2023 at 21:22:49 UTC, Richard (Rikki) Andrew 
Cattermole wrote:
 This use to work.

 It was removed a few years ago.

I think so too. I recently had to rework a lot of old code 
because of it, a real nuisance. Was it thought that null concat 
was dangerous somehow? A recipe for potential for some kind of 
mistakes? I presume that reinstating it would be safe enough? If 
there’s any reason to.

Dave P. <dave287091 gmail.com> writes:

On Sunday, 25 June 2023 at 21:30:52 UTC, Cecil Ward wrote:
 On Sunday, 25 June 2023 at 21:22:49 UTC, Richard (Rikki) Andrew 
 Cattermole wrote:
 This use to work.

 It was removed a few years ago.

 I think so too. I recently had to rework a lot of old code 
 because of it, a real nuisance. Was it thought that null concat 
 was dangerous somehow? A recipe for potential for some kind of 
 mistakes? I presume that reinstating it would be safe enough? 
 If there’s any reason to.

It’s easy to make this mistake:

```d
const x = [
      "foo"
      "bar",
      "baz"
];
```

`x` only has 2 elements, when it looks like it has 3.

Cecil Ward <cecil cecilward.com> writes:

On Sunday, 25 June 2023 at 21:39:13 UTC, Dave P. wrote:
 On Sunday, 25 June 2023 at 21:30:52 UTC, Cecil Ward wrote:
 On Sunday, 25 June 2023 at 21:22:49 UTC, Richard (Rikki) 
 Andrew Cattermole wrote:
 This use to work.

 It was removed a few years ago.

 I think so too. I recently had to rework a lot of old code 
 because of it, a real nuisance. Was it thought that null 
 concat was dangerous somehow? A recipe for potential for some 
 kind of mistakes? I presume that reinstating it would be safe 
 enough? If there’s any reason to.

 It’s easy to make this mistake:

 ```d
 const x = [
      "foo"
      "bar",
      "baz"
 ];
 ```

 `x` only has 2 elements, when it looks like it has 3.

An excellent point. I now fall in with the new status quo. :-)

Mathias LANG <geod24 gmail.com> writes:

On Sunday, 25 June 2023 at 21:47:01 UTC, Cecil Ward wrote:
 `x` only has 2 elements, when it looks like it has 3.

 An excellent point. I now fall in with the new status quo. :-)

And it's exactly the reasoning behind the deprecation.
You can find it in the changelog for 2.072, where this 
deprecation was introduced: 
https://dlang.org/changelog/2.072.0.html#deprecated_implicit_cat

Note that using the concat operator gives you exactly the same 
behavior: Concatenation happens at compile time, there is no 
runtime cost / GC allocation.