Re: [boost] [function] new implementation

Subject: Re: [boost] [function] new implementation
From: Lars Viklund (zao_at_[hidden])
Date: 2010-11-05 19:01:05

On Fri, Nov 05, 2010 at 10:54:17PM +0100, Domagoj Saric wrote:
>
> "Lars Viklund" <zao_at_[hidden]> wrote in message  
> news:20101104112534.GN22859_at_hatchepsut.acc.umu.se...
>> All tests that shall pass pass, and the two that shall fail fail on
>> 5.9 and 5.10 with r66387 and r66395.
>>
>> vacpp results are surprisingly good on r66395, log attached. It fails
>> just two tests unexpectedly, lib_function_test and function_n_test.
>
> Ah, the dreaded 'ampersandless' plain function pointer assignment...this  
> wrecks havoc on most compilers (including latest MSVC and Clang)...
> ...I've pretty much refactored the relevant code...try the new version 
> and let me know if it helped...
>
> ps. if you can, please also run tests for signals, signals2 and  
> program_options (these sometimes caught what the dedicated ones did 
> not)...

r66407 has the correct pass/failed-as-expected results on VisualAge.
All three test suites of program_options, signals and signals2 pass.

I ran the same three suites on Sun as well, just to cover all bases.

On 5.10, signals and signals2 pass completely, except for the usual
spurious mutex_test failure as indicated in the regression matrix.

program_options on 5.9 fails to compile parsers.o in a similar manner to
the one reported in the regular trunk test matrix, so it's most probably
not a Function fault (log attached).

Hooray!

-- 
Lars Viklund | zao_at_[hidden]