Re: [boost] Boost regression notification (2005-10-06 [RC_1_33_0])

From: Jonathan Wakely (cow_at_[hidden])
Date: 2005-10-07 18:22:01

Martin Wille wrote:

> We have an additional problem that does not become visible in the 
> reports Doug sends to this list. For gcc 4.0.2, there are a lot of 
> additional regressions. They look like a compiler problem to me, but I 
> didn't seriously investigate the problem.

Aren't some of those regressions only when _GLIBCXX_DEBUG is defined? 
It could be that the iterator-related tests are failing due to bugs in
the debug mode, but I've not looked into them yet.  Is GCC's debug mode
a release-critical platform ?

http://tinyurl.com/9eqr2
This crc-test is a compiler error, has anyone reported it?  If not I'll
add it to GCC's bugzilla.  This is a reduced version that triggers the
error:

#include <boost/crc.hpp>         // for boost::crc_basic, etc.

// Run tests on CRCs below a byte in size (here, 3 bits)
void small_crc_test1()
{
    unsigned char const  samples[4][4]
      = {
            { 0x3A, 0xC4, 0x08, 0x06 },
            { 0x42, 0xC5, 0x0A, 0x41 },
            { 0x4A, 0xC5, 0x08, 0x22 },
            { 0x52, 0xC4, 0x08, 0x05 }
        };

    boost::crc<3, 0x03, 0, 0, false, false>(samples[0], 4);
}

jon