Bug#959966: gdc -flto doesn't accept void main() signature.

[Iain Buclaw <ibuclaw@gdcproject.org>]

On 07/05/2020 17:16, Witold Baryluk wrote:
> Package: gdc
> Version: 4:9.2.1-3.1
> Severity: normal
> 
> Dear Maintainer,
> 
> According to language spec:
> 
> https://dlang.org/spec/function.html#main
> 
> """
> 19.20 main() Function
> 
> For console programs, main() serves as the entry point. It gets called
> after all the module initializers are run, and after any unittests are
> run. After it returns, all the module destructors are run. main() must be
> declared using one of the following forms:
> 
> void main() { ... }
> void main(string[] args) { ... }
> int main() { ... }
> int main(string[] args) { ... }
> """
> 
> 
> However:
> $ cat dlang.d
> void main() {
> }
> $ gdc -flto dlang.d
> /usr/lib/gcc/x86_64-linux-gnu/9/include/d/__entrypoint.di:33:5: warning: type of ‘_Dmain’ does not match original declaration [-Wlto-type-mismatch]
>    33 | int _Dmain(char[][] args);
>       |     ^
> dlang.d:2:6: note: type mismatch in parameter 1
>     2 | void main() {
>       |      ^
> dlang.d:2:6: note: type ‘void’ should match type ‘struct char[][]’
> dlang.d:2:6: note: ‘D main’ was previously declared here
> $
> 
> It does compile fine using `int mine(string[] args)` (and returning 0
> from it) and run correctly.
> 
> Also the message 'void' should match type 'struct char[][]' is confusing.
> 'struct'? void (return type?) should match it. Weird wording. I guess,
> maybe it is parameter 1, which is not in my definition, but should be
> string[]. But I am guessing here. Wording is weird and unclear what is
> what.
> 

Internally the last parameter type is 'void' to denote the end of the list (otherwise it's assumed the function is variadic).

Yes, it's a strange message, but this is coming from within the gcc middle-end, and warnings rarely make sense there post lowering.

This was fixed in PR d/90136, which is in gdc-10.

Iain.