Farid Zaripov wrote:
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
Sent: Saturday, March 24, 2007 1:10 AM
To: [email protected]
Subject: svn commit: r521934 -
/incubator/stdcxx/trunk/etc/config/src/STRUCT_TM_IN_WCHAR_H.cpp
Author: sebor
Date: Fri Mar 23 16:09:40 2007
New Revision: 521934
URL: http://svn.apache.org/viewvc?view=rev&rev=521934
Log:
2007-03-23 Martin Sebor <[EMAIL PROTECTED]>
[...]
-struct tm foo (struct tm arg)
+// 7.24.1, p2 of C99: struct tm is required to declared // as an
+incomplete type in <wchar.h>
+
+struct tm* struct_tm_declared (struct tm* arg)
{
return arg;
}
The my gcc 3.4.4 on cygwin and msvc treats the declaration of the
function
struct_tm_declared as the declaration of the type struct tm also.
I don't understand what you mean. Are you saying the compilers
issue an error for the test cases even when struct tm is declared
or that they do NOT issue an error when the struct is NOT declared?
If the former, can you post the error(s)?
In either case, it would help to have a test case for the compiler
bug (assuming it is, in fact, a bug).
I have reopened the issue STDCXX-364.
I propose to use using directive for checking the declaration of the
struct tm:
namespace {
using ::tm;
}
That relies on namespace support in the compiler which we don't
require (this goes back to the discussion about which compiler
features we should be allowed to assume; IIRC, we never decided
on a list).
Before we change how we test for struct tm I'd like to better
understand the problem with the test as it stands.
Martin
