suffix.d
```
void main ()
{
   real r = 1.L;
   float f = 1.f;
}
```

$ dmd suffix.d
suffix.d(3): Error: no property 'L' for type 'int'
suffix.d(4): Error: no property 'f' for type 'int'

According to the grammar in dmd2/html/d/spec/lex.html both are valid FloatLiterals. Any comments?

On Friday, 12 January 2018 at 12:45:59 UTC, kdevel wrote:
> suffix.d
> ```
> void main ()
> {
>    real r = 1.L;
>    float f = 1.f;
> }
> ```
>
> $ dmd suffix.d
> suffix.d(3): Error: no property 'L' for type 'int'
> suffix.d(4): Error: no property 'f' for type 'int'
>
> According to the grammar in dmd2/html/d/spec/lex.html both are valid FloatLiterals. Any comments?

Just found this on the same page

| If a floating literal has a . and a type suffix, at least one digit must be in-between:
|
| 1f; // OK
| 1.f; // forbidden
| 1.; // OK, double

Is there a rational for this restriction?

On Friday, 12 January 2018 at 12:57:37 UTC, kdevel wrote:
> On Friday, 12 January 2018 at 12:45:59 UTC, kdevel wrote:
>> suffix.d
>> ```
>> void main ()
>> {
>>    real r = 1.L;
>>    float f = 1.f;
>> }
>> ```
>>
>> $ dmd suffix.d
>> suffix.d(3): Error: no property 'L' for type 'int'
>> suffix.d(4): Error: no property 'f' for type 'int'
>>
>> According to the grammar in dmd2/html/d/spec/lex.html both are valid FloatLiterals. Any comments?
>
> Just found this on the same page
>
> | If a floating literal has a . and a type suffix, at least one digit must be in-between:
> |
> | 1f; // OK
> | 1.f; // forbidden
> | 1.; // OK, double
>
> Is there a rational for this restriction?

int foo(int n) { return n * 2; }

assert(2.foo == 4);

Now simply replace foo with f or L. I believe this ambiguity is the whole reason.

--
  Simen