Re: [Boost-testing] Regression testing failure on 'master'

Subject: Re: [Boost-testing] Regression testing failure on 'master'
From: Edward Diener (eldiener_at_[hidden])
Date: 2015-10-10 01:22:37

On 8/26/2015 12:09 PM, Jim Bell wrote:
>
> On 2015-08-25 2:10 PM, Edward Diener wrote:
>> ... the failure came after this:
>>
>>> ...
>>> ...found 1032 targets...
>>>
>>> c:\Programming\BoostRegression\boost_root\status>"c:\Programming\BoostRegression\boost_bb\src\engine\bin.ntx86\bjam.exe"
>>> "-sBOOST_BUILD_PATH=c:\Programming\BoostRegression;c:\Programming\BoostRegression\boost_bb\src"
>>> "-sBOOST_ROOT=c:\Programming\BoostRegression\boost_root"
>>> "--boost=c:\Programming\BoostRegression\boost_root"
>>> "--boost-root=c:\Programming\BoostRegression\boost_root"
>>> "--boost-build=c:\Programming\BoostRegression\boost_bb\src"
>>> "--debug-configuration"  -l300 toolset=gcc -d2
>>> preserve-test-targets=off --dump-tests  -m64
>>> "--build-dir=c:\Programming\BoostRegression\results"
>>> 1>>"c:\Programming\BoostRegression\results\bjam.log" 2>&1
>>>
>>> ...  (Usually occurs when bjam doesn't know how to make a target)
>>
>> Does anybody have any idea why the failure occurred ?
>>
>
> What does the tail end of results/bjam.log say?  Also, search it for
> always_show_run_output and timer/inspect.

Sorry for the very long delay in answering. I only recently found out 
that adding tests for the 'master' branch requires a PR to tell the test 
runner that these are new 'master' branch tests. That is the reason that 
the tests appeared to complete but never showed up on the 'master' 
regression report. For the 'develop' branch regression tests no such PR 
is required. I am going to work to add tests for mingw-64/gcc on 
Windows. Also if I have enough resources I would like to add tests for 
clang on Windows. For header-only libraries clang tests usually work 
very well but for non-header-only libraries clang has shown various type 
of link failures. So I do not know how successful clang regression 
testing on Windows will be.