Re: [V1][PATCH 1/3] Provide element_count attribute to flexible array member field (PR108896)

[Qing Zhao <qing.zhao@oracle.com>]

Hi,

Based on the discussion so far and further consideration, the following is my plan for this new attribute:

1.  The syntax of the new attribute will be:

__attribute__((counted_by (count_field_id)));

In the above, count_field_id is the identifier for the field that carries the number 
of elements info in the same structure of the FAM. 

For example:

struct object {
..
size_t count:  /* carries the number of elements info for the FAM flex.  */
int flex[] __attribute__((counted_by (count)));
};

2.  Later, if the argument of the this attribute need to be extended to an expression, we might need to 
extend the C FE to accept ".count”  in the future. 

Let me know if you have further comments and suggestions.

thanks.

Qing

> On Jun 20, 2023, at 3:40 PM, Qing Zhao via Gcc-patches <gcc-patches@gcc.gnu.org> wrote:
> 
> 
> 
>> On Jun 16, 2023, at 5:35 PM, Joseph Myers <joseph@codesourcery.com> wrote:
>> 
>> On Fri, 16 Jun 2023, Qing Zhao via Gcc-patches wrote:
>> 
>>>> So for 
>>>> 
>>>> struct foo { int c; int buf[(struct { int d; }){ .d = .c }]; };
>>>> 
>>>> one knows during parsing that the .d is a designator
>>>> and that .c is not.
>>> 
>>> Therefore, the above should be invalid based on this rule since .c is 
>>> not a member in the current structure.
>> 
>> What do you mean by "current structure"?  I think two different concepts 
>> are being conflated: the structure *being initialized* (what the C 
>> standard calls the "current object" for a brace-enclosed initializer 
>> list),
> 
> I think the concept of “current structure” should be stick to this. 
> 
>> and the structure *being defined*.
> Not this.
> 
> (Forgive me about my poor English -:)).
> 
> Then it will be cleaner? 
> 
> What’s your opinion?
> 
> 
>> The former is what's relevant 
>> for designators.  The latter is what's relevant for the suggested new 
>> syntax.  And .c *is* a member of the structure being defined in this 
>> example.
>> 
>> Those two structure types are always different, except for corner cases 
>> with C2x tag compatibility (where an object of structure type might be 
>> initialized in the middle of a redefinition of that type).
> 
> Can you give an example on this?  Thanks.
> 
> Qing
>> 
>> -- 
>> Joseph S. Myers
>> joseph@codesourcery.com