|
|
Boost Testing : |
Subject: [Boost-testing] Useless failures reports/broken testers
From: Gennadiy Rozental (rogeeff_at_[hidden])
Date: 2008-12-22 21:54:18
I've checked in some iffy code into Boost.Test and impatiently waiting
for regression results to appear here:
http://www.boost.org/development/tests/trunk/developer/test.html
to fix all regressions. What I see now is a bit unsettling (a lot
actually ;(). Number (majority) of testing setups/regression reports are
useless:
* Sandia-darwin-intel-11.0
Reports failure, but fail link leads to the message:
dyld: Library not loaded: libintlc.dylib
Referenced from:
/private/tmp/kbelco/boost/results/boost/bin.v2/libs/test/test/algorithms_test.test/intel-darwin-11.0/debug/algorithms_test
Reason: Incompatible library version: algorithms_test requires
version 1.0.0 or later, but libintlc.dylib provides version 0.0.0
* Sandia-darwin-intel-10.0
* HP-UX_ia64_gcc
* Sandia-pathscale
* Sandia-Linux-sun
* Sandia-intel-10.0
* CBIF-NetBSD-4.0-amd64
* CBIF-NetBSD-4.0-i386
* Sandia-sun in 2 incarnations (why are they not named differently BTW)
* siliconman (borland- 5.9.3)
* VeecoFTC (msvc-8.0~wm5~stlport5.1)
Reports failure in library build, but link to the library build output
shows success
* Huang-Vista-x64 (intel- vc9- win- 10.1_x86_64)
Reports failure in library build, but link to the library build output
shows nothing
* Huang-Vista-x64 (intel- vc9- win- 11.0_x86_64)
Reports failure in library build, but link to the library build output
shows:
call "D:/Program Files
(x86)/Intel/Compiler/11.0/066/cpp/Bin/Intel64//iclvars.bat" > nul
icl
@"F:\bv64\results\boost\bin.v2\libs\test\build\intel-vc9-win-11.0_x86_64\debug\address-model-64\threading-multi\compiler_log_formatter.obj.rsp"
'"D:/Program Files
(x86)/Intel/Compiler/11.0/066/cpp/Bin/Intel64//iclvars.bat"' is not
recognized as an internal or external command,
operable program or batch file.
'icl' is not recognized as an internal or external command,
operable program or batch file.
* Huang-WinXP-x86_32
Reports failure in library build, but link to the library build output
shows:
call "C:/Program
Files/Intel/Compiler/11.0/066/cpp/Bin/ia32//iclvars.bat" > nul
icl
@"G:\BoostRegressions\Head\results\boost\bin.v2\libs\test\build\intel-vc9-win-11.0\debug\threading-multi\compiler_log_formatter.obj.rsp"
'"C:/Program Files/Intel/Compiler/11.0/066/cpp/Bin/ia32//iclvars.bat"'
is not recognized as an internal or external command,
operable program or batch file.
'icl' is not recognized as an internal or external command,
operable program or batch file.
* siliconman (borland- 6.10.0)
Reports failure in library build, but link to the library build output
shows success with wierd content:
"C:/BuildAgent/work/1ca5584b164649a6/cb2009/bin/bcc32" -j5 -g255 -q
-c -P -Ve -Vx -a8 -b- -v -Od -w-8080 -tWC -tWR -tWC -WM-
-DBOOST_ALL_NO_LIB=1 -I".."
-I"C:/BuildAgent/work/1ca5584b164649a6/cb2009/include/"
-o"C:\BuildAgent\work\1ca5584b164649a6\results\boost\bin.v2\libs\test\build\borland-6.10.0\debug\link-static\test_main.obj"
"..\libs\test\src\test_main.cpp"
..\libs\test\src\test_main.cpp:
test_main.cpp is not supposed to be build at all. I do not use test
execution monitor any more.
* steven_watanabe-como
Reports:
call "C:/Program Files/Comeau/xp43101beta2/bin/setup"
como --no_version --no_prelink_verbose -o
"C:\regression\trunk\results\boost\bin.v2\libs\test\test\algorithms_test.test\como-win-4.3.10.1beta2\debug\link-static\algorithms_test"
@"C:\regression\trunk\results\boost\bin.v2\libs\test\test\algorithms_test.test\como-win-4.3.10.1beta2\debug\link-static\algorithms_test.exe.rsp"
Setting environment for using Microsoft Visual Studio 2008 x86 tools.
libboost_unit_test_framework-como43-d-1_38.lib(unit_test_parameters.obj)
: error LNK2005: std::basic_istream<T1, T2> &std::_M_get_num<T1, T2,
T3>(std::basic_istream<T1, T2> *, T3 &) [with T1=char,
T2=std::char_traits<char>, T3=unsigned int] already defined in
libboost_unit_test_framework-como43-d-1_38.lib(exception_safety.obj)
libboost_unit_test_framework-como43-d-1_38.lib(unit_test_parameters.obj)
: error LNK2005: T1 boost::detail::lexical_cast<T1, T2, N3,
T4>(boost::call_traits<T2>::param_type, T4 *, unsigned int) [with
T1=unsigned int, T2=boost::unit_test::basic_cstring<const char>,
N3=true, T4=char] already defined in
libboost_unit_test_framework-como43-d-1_38.lib(exception_safety.obj)
C:\regression\trunk\results\boost\bin.v2\libs\test\test\algorithms_test.test\como-win-4.3.10.1beta2\debug\link-static\algorithms_test
which essentially means that some of the STL symbols are multiply defined.
* Sandia-rh5-x86_64-intel
Reports:
Platform
Linux Redhat 5.1 Server (Tikanga) on 16 x Quad-Core Opteron
(Processor 8354), 1.1Ghz, 32Gb
Compilers
Intel(R) C Compiler for applications running on Intel(R) 64,
Version 10.1 Build 20080312 Package ID: l_cc_p_10.1.015
Options
-j64 -l300
Schedule
Full tests run nightly at 10:55pm, (Mountain Standard Time)
Contact
Noel Belcourt
kbelco at sandia.gov
Command Line
regression.py invoked as follows (trailing carets are line
continuation characters)
python regression.py --runner="Sandia-rh5-x86_64-intel"
--bjam-toolset=intel ^
--pjl-toolset=intel --toolsets="intel-10.1" --bjam-options="-j64
-l300"
Notes
User-config.jam
using python
: 2.4
;
using mpi
:
/apps/x86_64/mpi/openmpi/intel-10.1-f015-c015/openmpi-1.2.7_ofed/bin/mpic++
;
using intel
: 10.1
;
and that's it
* Sandia-intel-11.0
Reports:
Platform
Red Hat Enterprise Linux 2.6.9-34.0.2.ELsmp on Dual Duo-Core Intel
Xeon (3Ghz) with 8Gb RAM
Compilers
Intel(R) C++ Intel(R) 64 Compiler Professional for applications
running on Intel(R) 64, Version 11.0 Build 20081105 Package ID:
l_cproc_p_11.0.074
Options
-j8
Schedule
Full tests run nightly at 10:55pm, (Mountain Standard Time)
Contact
Noel Belcourt
kbelco at sandia.gov
Command Line
run.py invoked variously as follows
python run.py --runner="Sandia-intel-11.0"
--bjam-toolset=intel-linux-11.0 \
--pjl-toolset=intel-linux-11.0 --toolsets=intel-11.0
--bjam-options=-j8
and that's it
Can we fix these somehow?
Gennadiy