Subject: Re: [boost] [array] failures in c++11 mode
From: Marshall Clow (mclow.lists_at_[hidden])
Date: 2013-02-15 12:12:28
On Feb 15, 2013, at 7:39 AM, Marshall Clow <mclow.lists_at_[hidden]> wrote:
> On Feb 15, 2013, at 5:24 AM, John Maddock <boost.regex_at_[hidden]> wrote:
>>> I have built gcc 4.7.2 and run the array tests.
>>> I see the failure that you reported.
>>> I also see a gcc crash in the test "array_constexpr":
>>> darwin.compile.c++ ../../../bin.v2/libs/array/test/array_constexpr.test/darwin-11/debug/array_constexpr.o
>>> array_constexpr.cpp: In member function 'void test_main::test_method()':
>>> array_constexpr.cpp:32:36: in constexpr expansion of 'arr.boost::array<T, N>::at<int, 10ul>(3ul)'
>>> array_constexpr.cpp:32:36: internal compiler error: in adjust_temp_type, at cp/semantics.c:6400
>>> Please submit a full bug report,
>>> with preprocessed source if appropriate.
>>> See <http://gcc.gnu.org/bugs.html> for instructions.
>>> More as I find it out. ;-)
>> Just so you know, this is causing all the C++0x mode Boost.Math tests to fail currently.
>> Tricky one to solve though :-(
> I spoke to Johnathan Wakely about this yesterday, and he said:
>>>> I can't reproduce this crash.
>>> (Partially) reduced test case attached. (Still needs parts of boost)
>> Ah yes, that crashes with GCC 4.7.2, but not with the pre-release
>> 4.7.3 or with trunk - I'm pretty sure it's this bug I reported and the
>> code is valid:
> That's the good news.
> The bad news is that no released version of gcc compiles it.
Reverted the offending bits in 
Marshall Clow Idio Software <mailto:mclow.lists_at_[hidden]>
A.D. 1517: Martin Luther nails his 95 Theses to the church door and is promptly moderated down to (-1, Flamebait).
-- Yu Suzuki