On Sun, 21 Mar 2004 03:06:29 +0100, "Giovanni Bajo" <giovannibajo@libero.it> wrote:
Gennaro Prota wrote:
+ #elif defined(__MWERKS__) && (__MWERKS__ < 0x3003) \ + || (defined (__BORLANDC__) && \ + BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x564)))
This should really be:
#elif BOOST_WORKAROUND(__MWERKS__, < 0x3003) \ || BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x564))
Yes, it's just that the latter triggers gcc's -Wundef. I don't have a strong opinion here, but there seems to be agreement that gcc users should be allowed to use that switch. If you are wondering why BOOST_WORKAROUND isn't itself defined in terms of the defined operator: http://lists.boost.org/MailArchives/boost/msg40249.php
And, didnt't we settle for using BOOST_MY_COMPILER instead of checking __MY_COMPILER__ directly?
Well, yes, we could wait for the config system to be updated, before changing static_assert.hpp.
We could try and follow this convention at least for new code we add.
Agreed.
Secondly, is it ok that 'STATIC_ASSERTION_FAILURE' is all-uppercase but not prefixed by 'BOOST_'?
Isn't that a class? Then it doesn't matter.
<paranoid> But it can conflict with user defined macros. </paranoid>
BTW, "agurt" is the (nick)name of whoever added that comment line. cvs annotate is your friend if you can't unmangle it.
Aah! How could I miss that. Sorry, Aleksey! -- Genny.