'$' is only valid in an indexExpression (https://dlang.org/spec/grammar.html#IndexExpression),
so it can only be followed by

- ' '
- ']'
-  operators , usually '-' but also '/', '+', '>>' etc

Is that right ?

I'd like to relax the lexical rule for C.E static macros which currently is

- "^\$\w*[a-zA-Z]$", so for example "$a1A" is valid and "$a1" is not.

But it looks like for example "$)" or "$}" wouldn't be ambiguous since it's not possible in D.

On Sunday, 13 March 2016 at 14:07:31 UTC, Basile B. wrote:
> '$' is only valid in an indexExpression (https://dlang.org/spec/grammar.html#IndexExpression),
> so it can only be followed by
>
> - ' '
> - ']'
> -  operators , usually '-' but also '/', '+', '>>' etc
>
> Is that right ?
>
> I'd like to relax the lexical rule for C.E static macros which currently is
>
> - "^\$\w*[a-zA-Z]$", so for example "$a1A" is valid and "$a1" is not.
>
> But it looks like for example "$)" or "$}" wouldn't be ambiguous since it's not possible in D.

I don't know what C.E is, but `$` is an expansion of `PrimaryExpression`, which means it can (syntactically) appear anywhere a `PrimaryExpression` is allowed. For example, this compiles:

    void main() {
        int[] a;
        a[0 .. ($)] = 0;
    }

On Sunday, 13 March 2016 at 14:55:36 UTC, Marc Schütz wrote:
> On Sunday, 13 March 2016 at 14:07:31 UTC, Basile B. wrote:
>> '$' is only valid in an indexExpression (https://dlang.org/spec/grammar.html#IndexExpression),
>> so it can only be followed by
>>
>> - ' '
>> - ']'
>> -  operators , usually '-' but also '/', '+', '>>' etc
>>
>> Is that right ?
>>
>> I'd like to relax the lexical rule for C.E static macros which currently is
>>
>> - "^\$\w*[a-zA-Z]$", so for example "$a1A" is valid and "$a1" is not.
>>
>> But it looks like for example "$)" or "$}" wouldn't be ambiguous since it's not possible in D.
>
> I don't know what C.E is, but `$` is an expansion of `PrimaryExpression`, which means it can (syntactically) appear anywhere a `PrimaryExpression` is allowed. For example, this compiles:
>
>     void main() {
>         int[] a;
>         a[0 .. ($)] = 0;
>     }

Then the C.E macros syntax can't be enhanced. At least the current is not ambiguous.

The thing is that a macro is substitued automatically/dynamically so the requirement is that nothing that's syntaxically valid in D must replacable.