On 7/22/25 12:46, juan via agora-discussion wrote:
> juan via agora-discussion [2025-07-22 13:21]:
>> Janet Cobb via agora-discussion [2025-07-21 19:32]:
>>> On 7/21/25 19:30, Mischief via agora-discussion wrote:
>>>> On 7/21/25 12:09 PM, juan via agora-business wrote:
>>>>> Cosmo via agora-official [2025-07-20 23:29]:
>>>>>> I initiate a referendum on each of the following proposals, removing
>>>>>> them from the proposal pool. […]
>>>>> I do the actions specified in the document that is formed by the output
>>>>> of the following C program as compiled by the ANSI standard:
>>>>>
>>>>> {
>>>>> main(){char i;for (i = 0; i < 7; i++){printf("I vote FOR %d.\n", 9234 +
>>>>> i);}}
>>>>> }
>>>> P9240 hasn't been distributed yet.
>>>>
>>>> (Also, gcc wouldn't compile this as-is for me since it wanted cstdio
>>>> #include'd.)
>>>>
>>> Yes, under the standard, this is ill-formed. I think those votes fail.
>> […]
> Further investigation yielded the same conclusion. Under the C89 standard,
> function calls composed of a single identifier that has no declaration act
> as if declared as returning an `int`. This is the case for both `main` and
> `printf`. See this link: https://stackoverflow.com/a/437763/11672072. I
> checked in the actual spec, and the quote is correct.
>
> I suggest verifying your compiler's compliance to spec.
>
> Note: the spec itself gives as an example calling a library function
> without declaring it (in §7.1.7, the last example).
>
Yes, sorry. I was wrong above. It's not "ill-formed" (which is C++
terminology; in C, the equivalent would be a constraint violation); it's
just UB. See my arguments for the CFJ I just called.
--
Janet Cobb
Assessor, Rulekeepor
