[Bug c/114266] New: No -pedantic diagnostic for zero-sized array in compound literals

[Bug c/114266] New: No -pedantic diagnostic for zero-sized array in compound literals

[daniel.lundin.mail at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114266

            Bug ID: 114266
           Summary: No -pedantic diagnostic for zero-sized array in
                    compound literals
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: daniel.lundin.mail at gmail dot com
  Target Milestone: ---

(int[]){}; produces no diagnostics with -std=c23 -pedantic.

Empty initalizer lists are OK under -std=c23 and -std=gnu so that's not the
problem. gcc correctly gives a diagnostic for that in -std=c17 -pedantic mode.

Zero-sized arrays should however give a diagnostic in -std=cxx -pedantic mode.
Prior to C23 this was not really a problem as we would get a diagnostic for the
empty initializer list. But now this non-conforming C compiles cleanly even
with -pedantic set.

A diagnostic similar to the one given in clang would be good:
warning: zero size arrays are an extension [-Wzero-length-array]

As I understand it, gcc did have -Wzero-length-array in the past but removed
it? See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94428. The options
-Wstrict-flex-arrays -fstrict-flex-arrays seem to have no effect (and my
example isn't a flexible array member but a compound literal).

[Bug c/114266] No -pedantic diagnostic for zero-sized array in compound literals

[daniel.lundin.mail at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114266

--- Comment #1 from Daniel Lundin <daniel.lundin.mail at gmail dot com> ---
Created attachment 57643
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=57643&action=edit
Complete example

[Bug c/114266] No -pedantic diagnostic for zero-sized array in compound literals

[jsm28 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114266

Joseph S. Myers <jsm28 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2024-03-07

--- Comment #2 from Joseph S. Myers <jsm28 at gcc dot gnu.org> ---
The relevant constraint here is "An array of unknown size shall not be
initialized by an empty initializer.".

[Bug c/114266] No -pedantic diagnostic for zero-sized array in compound literals

[daniel.lundin.mail at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114266

--- Comment #3 from Daniel Lundin <daniel.lundin.mail at gmail dot com> ---
(In reply to Joseph S. Myers from comment #2)
> The relevant constraint here is "An array of unknown size shall not be
> initialized by an empty initializer.".

Indeed! I didn't realize it was also a constraint (C23 6.7.10). That probably
means that a diagnostic warning should always be given and not just in
-pedantic mode.