Gennaro Prota wrote:
On Sun, 23 Jul 2006 21:37:03 +0200, Gennaro Prota <gennaro_prota@yahoo.com> wrote:
Hi guys,
among the various Sun CC failures that permeate our regression tests there's one which seems quite straightforward to fix [...]
Another easy one:
in "../boost/iostreams/char_traits.hpp", line 40 and others: wint_t is not defined. The file char_traits.hpp tries bringing that name into std twice: first via the BOOST_NO_STDC_NAMESPACE and then via an ad-hoc #if (which should actually use BOOST_WORKAROUND).
I'll have to check this but I think some of the changes that Simon Atanasyan from SUN made and published in his blog address this. It's a tad late in the day for me to check but it may well be that I've got a patch for that which I'm preparing...
Now I'm not sure whether we should define BOOST_NO_STDC_NAMESPACE when __SUNPRO_CC == 0x580 (probably not, as the problem might only exist for rarely used names such as wint_t) or add to the second #if. In both cases that's another easy fix.
I don't think we should as the compiler in general is fairly standard compliant. However SUN mainly supports the configuration of this compiler + STLport. If you don't use that configuration you may be seeing some really odd things happening. Regards, Timo