Eric Lemings wrote:
-----Original Message-----
From: Eric Lemings
Sent: Monday, April 21, 2008 9:43 AM
To: Martin Sebor; Travis Vitek
Subject: RE: //stdcxx/trunk/tests/localization/22_synopsis.cpp
-----Original Message-----
From: Martin Sebor [mailto:[EMAIL PROTECTED]
Sent: Friday, April 18, 2008 2:20 PM
To: Travis Vitek
Cc: Eric Lemings
Subject: Re: //stdcxx/trunk/tests/localization/22_synopsis.cpp
Travis Vitek wrote:
Eric Lemings wrote:
In the array initialization of lc_defs, there's a null string
before "LC_MESSAGES". Is that a typo?
It won't hurt anything. That code is there to make sure
that the LC_*
constants aren't pulled in when including <locale>. The
only assertion
that uses the array just iterates over it asserting for
non-null values.
I'd probably opt to remove the null, but that's just me.
I'm actually quite interested to know why that test does this...
#if defined (_RWSTD_NO_EXPORT) && !defined (_MSC_VER)
// make protected members accessible if possible
# define protected public
#endif
That is just craziness.
http://perforce.cvo.roguewave.com/@//@/165436?ac=10
Martin
Why is this dependent on whether _RWSTD_NO_EXPORT is defined?
Because the only compiler that implements the feature (exported
templates), EDG eccp, detects the ODR violation in the hack above
and gives an error.
I'm testing with gcc on Linux and this is not defined which
in turn leads to compile errors.
And where is get_locale_category()? I can't find it anywhere.
See the %{Lc} directive to rw_printf(). Here are some use cases:
http://fisheye6.cenqua.com/browse/stdcxx/trunk/tests/self/0.printf.cpp?r=trunk#l1061
Martin
