[Bug libstdc++/99270] New: Testsuite 27_io/headers/cstdio/types_std.cc failure on DragonFly

[Bug libstdc++/99270] New: Testsuite 27_io/headers/cstdio/types_std.cc failure on DragonFly

[rimvydas.jas at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99270

            Bug ID: 99270
           Summary: Testsuite 27_io/headers/cstdio/types_std.cc failure on
                    DragonFly
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libstdc++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rimvydas.jas at gmail dot com
  Target Milestone: ---

Testsuite on x86_64-*-dragonfly gives:

Running target unix
FAIL: 27_io/headers/cstdio/types_std.cc (test for excess errors)

spawn -ignore SIGHUP /build/trunk/./gcc/xg++ -shared-libgcc
-B/build/trunk/./gcc -nostdinc++
-L/build/trunk/x86_64-unknown-dragonfly5.9/libstdc++-v3/src
-L/build/trunk/x86_64-unknown-dragonfly5.9/libstdc++-v3/src/.libs
-L/build/trunk/x86_64-unknown-dragonfly5.9/libstdc++-v3/libsupc++/.libs
-B/opt/gcc11f/x86_64-unknown-dragonfly5.9/bin/
-B/opt/gcc11f/x86_64-unknown-dragonfly5.9/lib/ -isystem
/opt/gcc11f/x86_64-unknown-dragonfly5.9/include -isystem
/opt/gcc11f/x86_64-unknown-dragonfly5.9/sys-include -fchecking=1
-B/build/trunk/x86_64-unknown-dragonfly5.9/./libstdc++-v3/src/.libs
-fmessage-length=0 -fno-show-column -ffunction-sections -fdata-sections -g -O2
-DLOCALEDIR="." -nostdinc++
-I/build/trunk/x86_64-unknown-dragonfly5.9/libstdc++-v3/include/x86_64-unknown-dragonfly5.9
-I/build/trunk/x86_64-unknown-dragonfly5.9/libstdc++-v3/include
-I/data/gg/libstdc++-v3/libsupc++ -I/data/gg/libstdc++-v3/include/backward
-I/data/gg/libstdc++-v3/testsuite/util
/data/gg/libstdc++-v3/testsuite/27_io/headers/cstdio/types_std.cc
-fdiagnostics-plain-output -S -o types_std.s
/data/gg/libstdc++-v3/testsuite/27_io/headers/cstdio/types_std.cc:25: error:
aggregate 'FILE gnu::f' has incomplete type and cannot be defined
compiler exited with status 1
FAIL: 27_io/headers/cstdio/types_std.cc (test for excess errors)
Excess errors:
/data/gg/libstdc++-v3/testsuite/27_io/headers/cstdio/types_std.cc:25: error:
aggregate 'FILE gnu::f' has incomplete type and cannot be defined

extra_tool_flags are:

On DragonFly the FILE type is declared as opaque struct and its internals are
never exposed on purpose:
typedef struct __FILE FILE;

Even cppreference mentions that std::FILE may be semantically non-copyable.
Could "std::FILE f;" test be excluded by #if !defined(__DragonFly__) or simply
changed to a more usual use case of "std::FILE *fp;" ?

[Bug libstdc++/99270] Testsuite 27_io/headers/cstdio/types_std.cc failure on DragonFly

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99270

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2021-02-25
     Ever confirmed|0                           |1

--- Comment #1 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Rimvydas (RJ) from comment #0)
> simply changed to a more usual use case of "std::FILE *fp;" ?

Yes, I think that, or declaring extern std::FILE f; instead.

[Bug libstdc++/99270] Testsuite 27_io/headers/cstdio/types_std.cc failure on DragonFly

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99270

--- Comment #2 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Interestingly, in C99 we have:

  Types are partitioned into object types (types that fully describe objects),
  function types (types that describe functions), and incomplete types (types
  that describe objects but lack information needed to determine their sizes).

And FILE is specified to be an object type, which means it can't be incomplete.

http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1439.pdf changed that text for
C11 so that incomplete types are a sub-category of object type, and
intentionally didn't change the definition of FILE, allowing it to be
incomplete (see last page of N1439).

Not that it matters, we should fix the test either way. Just a curiosity.

[Bug libstdc++/99270] Testsuite 27_io/headers/cstdio/types_std.cc failure on DragonFly

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99270

--- Comment #3 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jonathan Wakely <redi@gcc.gnu.org>:

https://gcc.gnu.org/g:31002af9040876dfb6444fabd524b11ca0739519

commit r11-7391-g31002af9040876dfb6444fabd524b11ca0739519
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Thu Feb 25 15:31:19 2021 +0000

    libstdc++: Do not assume std::FILE is complete [PR 99270]

    libstdc++-v3/ChangeLog:

            PR libstdc++/99270
            * testsuite/27_io/headers/cstdio/types_std.cc: Use pointer to
            FILE instead of FILE.

[Bug libstdc++/99270] Testsuite 27_io/headers/cstdio/types_std.cc failure on DragonFly

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99270

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Thanks