[cfe-dev] Duplicate Qualifiers from AST

[Eli Friedman]

On Fri, Aug 5, 2011 at 12:07 PM, Philippe Hausler <phausler at gmail.com> wrote:
> I am not sure if this is a bug or a very arcane C standard: perhaps some clarification is needed.
>
> How is it syntactically correct when parsing C to have a duplicate qualifiers like const or static on a primitive type like int or char?
>
> Example:
>
> const const static static unsigned unsigned int foo = 0;
>
> This behavior is exhibited in both the AST as well as in higher level usages e.g. Xcode's syntax highlighter.
>
> Granted the attributes are not conflicting, but should this not generate a warning (which it seems that clang does not have such a warning for duplicate qualifiers applied as such). However I understand the reason why this does not generate an error since there is no inherit conflict between the qualifier list.
>
> It is understandable that a pointer-pointer could potentially have two const qualifiers; one to determine the first indirection's "const-ness" and another to determine the second indirection's "const-ness". Is there some other reasoning similar for primitive types to have this as well, or have I just missed the boat on some flag somewhere?
>
> As a side note GCC pitches a fit about the duplication of static or unsigned but not a peep about the double const.

C99 6.7.3p4 explicitly states that double-const is legal (and
equivalent to a single const).  The other two will generate warnings
if you pass -pedantic to clang.  I don't know off the top of my head
why that warning isn't on by default.

-Eli