foreach(auto v; msg)
  writeln(v);

gives an error that a basic type is expected

foreach(v; msg)
  writeln(v);

works

.. but why?

On Saturday, 16 July 2016 at 14:00:56 UTC, dom wrote:
> foreach(auto v; msg)
>   writeln(v);
>
> gives an error that a basic type is expected
>
> foreach(v; msg)
>   writeln(v);
>
> works
>
> .. but why?


`Note: The ForeachTypeAttribute is implicit, and when a type is not specified, it is inferred. In that case, auto is implied, and it is not necessary (and actually forbidden) to use it.`
http://dlang.org/spec/statement.html#ForeachStatement

On Saturday, 16 July 2016 at 14:00:56 UTC, dom wrote:
> foreach(auto v; msg)
>   writeln(v);
>
> gives an error that a basic type is expected
>
> foreach(v; msg)
>   writeln(v);
>
> works
>
> .. but why?

Arbitrary limitation. If you want to say how surprising and uselessly limiting it is wait at the end of the line.

It's not actually a problem in practice because you just have not to put it but it is part of those frustrating little edge cases with no reason to be.

On Saturday, 16 July 2016 at 14:00:56 UTC, dom wrote:
> .. but why?

because. i've lost that fight too.

On Saturday, 16 July 2016 at 14:11:34 UTC, cym13 wrote:
> On Saturday, 16 July 2016 at 14:00:56 UTC, dom wrote:
>> foreach(auto v; msg)
>>   writeln(v);
>>
>> gives an error that a basic type is expected
>>
>> foreach(v; msg)
>>   writeln(v);
>>
>> works
>>
>> .. but why?
>
> Arbitrary limitation. If you want to say how surprising and uselessly limiting it is wait at the end of the line.
>
> It's not actually a problem in practice because you just have not to put it but it is part of those frustrating little edge cases with no reason to be.

It's not arbitrary. It keeps the language simple and easy to read. After all the entire auto keyword is just there, because the compiler needs a keyword and in loops it's clearly defined what the type will be.
You don't complain that `int int` is forbidden, or do you?

I guess you are frustrated because you are used to this pattern from other, inferior languages. I bet you will soon start to appreciate the syntactic sugar that D provides.

On Saturday, 16 July 2016 at 20:00:39 UTC, Seb wrote:
> On Saturday, 16 July 2016 at 14:11:34 UTC, cym13 wrote:
>> [...]
>
> It's not arbitrary. It keeps the language simple and easy to read. After all the entire auto keyword is just there, because the compiler needs a keyword and in loops it's clearly defined what the type will be.
> You don't complain that `int int` is forbidden, or do you?
>
> I guess you are frustrated because you are used to this pattern from other, inferior languages. I bet you will soon start to appreciate the syntactic sugar that D provides.

However I agree that the error message should be more informative. Please open a issue for this ;-)

On Saturday, 16 July 2016 at 20:00:39 UTC, Seb wrote:
> On Saturday, 16 July 2016 at 14:11:34 UTC, cym13 wrote:
>> On Saturday, 16 July 2016 at 14:00:56 UTC, dom wrote:
>>> foreach(auto v; msg)
>>>   writeln(v);
>>>
>>> gives an error that a basic type is expected
>>>
>>> foreach(v; msg)
>>>   writeln(v);
>>>
>>> works
>>>
>>> .. but why?
>>
>> Arbitrary limitation. If you want to say how surprising and uselessly limiting it is wait at the end of the line.
>>
>> It's not actually a problem in practice because you just have not to put it but it is part of those frustrating little edge cases with no reason to be.
>
> It's not arbitrary. It keeps the language simple and easy to read. After all the entire auto keyword is just there, because the compiler needs a keyword and in loops it's clearly defined what the type will be.
> You don't complain that `int int` is forbidden, or do you?
>
> I guess you are frustrated because you are used to this pattern from other, inferior languages. I bet you will soon start to appreciate the syntactic sugar that D provides.

int int is a completely different issue, and that has nothing to do with what you suppose my experience is from other languages. If anything I come mainly from python so I find not having to write the type a pretty thing.

However auto should be allowed here. You are defining a variable and the fact that it's in a foreach shouldn't be of any importance. The language should enforce orthogonality of orthogonal things, not break it. A variable definition in a foreach should be allowed not because I like to be pedantic but because it's allowed everywhere else. That also stands for the principle of least surprise.

On Saturday, 16 July 2016 at 21:39:42 UTC, cym13 wrote:
> A variable definition in a foreach

-> A variable definition *with auto* in a foreach

On Saturday, 16 July 2016 at 21:39:42 UTC, cym13 wrote:
> However auto should be allowed here. You are defining a variable and the fact that it's in a foreach shouldn't be of any importance. The language should enforce orthogonality of orthogonal things, not break it. A variable definition in a foreach should be allowed not because I like to be pedantic but because it's allowed everywhere else. That also stands for the principle of least surprise.

actually, `foreach (v; rng)` looks like `foreach` is *reusing* *existing* *variable*. most of the time you can put `immutable` or something like that there to note that it is not reusing (purely cosmetical thing), but sometimes you cannot, and then `auto` is perfect candidate... but it is not allowed. (sigh)

On Saturday, 16 July 2016 at 22:05:49 UTC, ketmar wrote:
> actually, `foreach (v; rng)` looks like `foreach` is *reusing* *existing* *variable*. most of the time you can put `immutable` or something like that there to note that it is not reusing (purely cosmetical thing), but sometimes you cannot, and then `auto` is perfect candidate... but it is not allowed. (sigh)

Chipping in my agreement. foreach(x; y) makes as much syntactic sense as for(x = 0; x < y; x++) where x was not previously defined. One does not expect something that does not look like every other variable definition in the language to be defining a new variable.