Re: [boost] Regression test woes

2002-12-10 Thread Beman Dawes
At 04:58 PM 12/9/2002, David Abrahams wrote:

Beman Dawes [EMAIL PROTECTED] writes:

 First, the regression tests are broken; the message is:

 unknown dependent target
libs!test!buildlibboost_prg_exec_monitor.lib

 I assume this is the result of a change Glennadiy made last night in
 the test library build Jamfile, but reverting it to the prior version
 doesn't clear the problem. Sigh. I've emailed him privately.

I fixed up status/Jamfile to refer to the new targets.

Thanks! The bjam part is now running, but the post test processing is 
failing. The failure is occurring in the Boost.test tests library. I won't 
be able to look at it until late afternoon; the error message from the 
post-processing program was less that helpful.

We need to find a way to better test changes to the test library because 
failure in it cause the whole regression testing system to fail. Perhaps 
all changes to Boost.Test should first go on a branch, and only be merged 
when they have been tested with several compilers and operating systems.

--Beman


___
Unsubscribe  other changes: http://lists.boost.org/mailman/listinfo.cgi/boost


RE: [boost] Regression test woes

2002-12-10 Thread Beman Dawes
At 05:47 PM 12/9/2002, Aleksey Gurtovoy wrote:

David Abrahams wrote:
  Second, with VC++ large numbers of warnings are now
  appearing for many type traits headers.

 Which version? I can't reproduce these problems with 6, 7, or 7.1.

I can't either, but I fixed it anyway :).

Thanks! They went away for me, too.

--Beman


___
Unsubscribe  other changes: http://lists.boost.org/mailman/listinfo.cgi/boost



RE: [boost] Regression test woes

2002-12-10 Thread Rozental, Gennadiy
 Thanks! The bjam part is now running, but the post test processing is 
 failing. The failure is occurring in the Boost.test tests 
 library. I won't 
 be able to look at it until late afternoon; the error message 
 from the 
 post-processing program was less that helpful.

You confusing me again. What is post test processing? And how it related
to Boost.Test tests?
Also AFAIK status/Jamfile does not have them yet (about to be added)
 
 We need to find a way to better test changes to the test 
 library because 
 failure in it cause the whole regression testing system to 
 fail. Perhaps 
 all changes to Boost.Test should first go on a branch, and 
 only be merged 
 when they have been tested with several compilers and 
 operating systems.

I am trying to get a naccess to as much compilers as I could. At least
windows compilers. I do not commit the changes until all my tests are
passing.

Gennadiy.
___
Unsubscribe  other changes: http://lists.boost.org/mailman/listinfo.cgi/boost



RE: [boost] Regression test woes

2002-12-09 Thread Rozental, Gennadiy
 First, the regression tests are broken; the message is:
 
 unknown dependent target 
 libs!test!buildlibboost_prg_exec_monitor.lib

1. I do not get it. boost_prg_exec_monitor is the new name of the library.
bjam could not find it?
2. Where did you get these errors?
3. I though that in most cases minimal testing is used. So majority of boost
regression tests should ignore library rename. And only couple of libraries
that does use external library will need to change the name in Jamfile.

 
 I assume this is the result of a change Glennadiy made last 
 night in the 
 test library build Jamfile, but reverting it to the prior 
 version doesn't 
 clear the problem. Sigh. I've emailed him privately.

I did not get it. Could you resend?

Gennadiy.
___
Unsubscribe  other changes: http://lists.boost.org/mailman/listinfo.cgi/boost



Re: [boost] Regression test woes

2002-12-09 Thread David Abrahams
Beman Dawes [EMAIL PROTECTED] writes:

 First, the regression tests are broken; the message is:

 unknown dependent target libs!test!buildlibboost_prg_exec_monitor.lib

 I assume this is the result of a change Glennadiy made last night in
 the test library build Jamfile, but reverting it to the prior version
 doesn't clear the problem. Sigh. I've emailed him privately.

I fixed up status/Jamfile to refer to the new targets.

I also removed the explicit use of $(SUFLIB) that people keep
stubbornly putting in their dependency references despite the fact
that I keep telling them it's just unneeded clutter :-/

 Second, with VC++ large numbers of warnings are now appearing for many
 type traits headers.  

Which version? I can't reproduce these problems with 6, 7, or 7.1.

However CWPro8.3 shows lots of trivial bugs in the test library:

metrowerks-C++-action 
c:\build\libs\test\build\bin\libboost_unit_test_framework.lib\cwpro8.3\debug\runtime-link-dynamic\execution_monitor.obj
### mwcc Compiler:
#File: ..\libs\test\src\execution_monitor.cpp
# ---
# 157:  case _CRT_ASSERT:
#   Error:   ^^^
#   undefined identifier '_CRT_ASSERT'
### mwcc Compiler:
# 161:  case _CRT_ERROR:
#   Error:   ^^
#   undefined identifier '_CRT_ERROR'

Errors caused tool to abort.

call c:\tools\cwpro8.3\Other Metrowerks Tools\Command Line Tools\cwenv.bat 
-quiet 
mwcc  -maxerrors 5 -maxwarnings 20 -c -warn on,nounusedexpr,nounused -cwd include  
 -DNOMINMAX -nowraplines -lang c++ -g -O0 -inline off -prefix UseDLLPrefix.h  
-iso_templates on -fl opt_classresults   -Ic:\build\libs\test\build  -I- 
-Ic:\boost  -o 
c:\build\libs\test\build\bin\libboost_unit_test_framework.lib\cwpro8.3\debug\runtime-link-dynamic\execution_monitor.obj
  ..\libs\test\build\../src\execution_monitor.cpp 

...failed metrowerks-C++-action 
c:\build\libs\test\build\bin\libboost_unit_test_framework.lib\cwpro8.3\debug\runtime-link-dynamic\execution_monitor.obj...
metrowerks-C++-action 
c:\build\libs\test\build\bin\libboost_unit_test_framework.lib\cwpro8.3\debug\runtime-link-dynamic\test_tools.obj
### mwcc Compiler:
#  In: ..\boost\test\unit_test_result.hpp
#From: ..\libs\test\src\test_tools.cpp
# 
#  53:  voidshort_report( std::ostream where_to_, size_t indent_ 
= 0 );// short
#   Error: ^^
#   declaration syntax error
### mwcc Compiler:
#  54:  voiddetailed_report( std::ostream where_to_, size_t 
indent_ = 0 ); // long
#   Error:^^
#   declaration syntax error
### mwcc Compiler:
#  70:  private:
#   Error:  ^^^
#   declaration syntax error
### mwcc Compiler:
#  75:  unit_test_result( unit_test_result* parent_, c_string_literal 
test_case_name_, unit_test_counter expected_failures_ = 0 )
#   Error:^
#   ')' expected
### mwcc Compiler:
#  79:  };
# Warning:   ^
#   illegal empty declaration
### mwcc Compiler:
#  79:  };
# Warning:   ^
#   illegal empty declaration
### mwcc Compiler:
#  93:  } // namespace boost
#   Error:  ^
#   declaration syntax error
#   Too many errors printed, aborting program

User break, cancelled...

call c:\tools\cwpro8.3\Other Metrowerks Tools\Command Line Tools\cwenv.bat 
-quiet 
mwcc  -maxerrors 5 -maxwarnings 20 -c -warn on,nounusedexpr,nounused -cwd include  
 -DNOMINMAX -nowraplines -lang c++ -g -O0 -inline off -prefix UseDLLPrefix.h  
-iso_templates on -fl opt_classresults   -Ic:\build\libs\test\build  -I- 
-Ic:\boost  -o 
c:\build\libs\test\build\bin\libboost_unit_test_framework.lib\cwpro8.3\debug\runtime-link-dynamic\test_tools.obj
  ..\libs\test\build\../src\test_tools.cpp 

...failed metrowerks-C++-action 
c:\build\libs\test\build\bin\libboost_unit_test_framework.lib\cwpro8.3\debug\runtime-link-dynamic\test_tools.obj...
metrowerks-C++-action 
c:\build\libs\test\build\bin\libboost_unit_test_framework.lib\cwpro8.3\debug\runtime-link-dynamic\unit_test_parameters.obj
metrowerks-C++-action 
c:\build\libs\test\build\bin\libboost_unit_test_framework.lib\cwpro8.3\debug\runtime-link-dynamic\unit_test_log.obj
### mwcc Compiler:
#  In: ..\boost\test\unit_test_result.hpp
#From: ..\libs\test\src\unit_test_log.cpp
# ---
#  53:  voidshort_report( std::ostream where_to_, size_t indent_ 
= 0 );// short
#   Error: ^^
#   declaration syntax error
### mwcc Compiler:
#  54:  voiddetailed_report( std::ostream where_to_, size_t 
indent_ = 0 ); // long
#   Error:^^
#   declaration syntax error
### mwcc Compiler:
#  70:  private:
# 

Re: [boost] Regression test woes

2002-12-09 Thread David Abrahams
Rozental, Gennadiy [EMAIL PROTECTED] writes:

 First, the regression tests are broken; the message is:
 
 unknown dependent target 
 libs!test!buildlibboost_prg_exec_monitor.lib

 1. I do not get it. boost_prg_exec_monitor is the new name of the library.
 bjam could not find it?

I think he mis-copied. Mine said something like:

 unknown dependent target libs!test!buildlibprg_exec_monitor.lib

 2. Where did you get these errors?

Umm, from running bjam test in the status directory. Where else do
boost regression test results come from?

 3. I though that in most cases minimal testing is used. So majority of boost
 regression tests should ignore library rename. And only couple of libraries
 that does use external library will need to change the name in
 Jamfile.

Why don't you just do a grep through the boost sources for your old
library names? It seems as though, if you're going to change the
library names, it should be easy enough to fix up all the Jamfiles to
work with the new ones.

-- 
   David Abrahams
   [EMAIL PROTECTED] * http://www.boost-consulting.com
Boost support, enhancements, training, and commercial distribution

___
Unsubscribe  other changes: http://lists.boost.org/mailman/listinfo.cgi/boost



RE: [boost] Regression test woes

2002-12-09 Thread Aleksey Gurtovoy
David Abrahams wrote:
  Second, with VC++ large numbers of warnings are now 
  appearing for many type traits headers.  
 
 Which version? I can't reproduce these problems with 6, 7, or 7.1.

I can't either, but I fixed it anyway :).

Aleksey
___
Unsubscribe  other changes: http://lists.boost.org/mailman/listinfo.cgi/boost



RE: [boost] Regression test woes

2002-12-09 Thread Rozental, Gennadiy
 #File: ..\libs\test\src\execution_monitor.cpp
 # ---
 # 157:  case _CRT_ASSERT:
 #   Error:   ^^^
 #   undefined identifier '_CRT_ASSERT'
 ### mwcc Compiler:
 # 161:  case _CRT_ERROR:
 #   Error:   ^^
 #   undefined identifier '_CRT_ERROR'

What kind of C runtime hooks metrowerks supports?

 
 Errors caused tool to abort.
 
 call c:\tools\cwpro8.3\Other Metrowerks Tools\Command 
 Line Tools\cwenv.bat -quiet 

Could I get these command line tools somewhere to install on my local PC for
testing of Boost.Test?

 #From: ..\libs\test\src\test_tools.cpp
 # 
 #  53:  voidshort_report( std::ostream where_to_,
size_t indent_ = 0 );// short
 #   Error:
^^
 #   declaration syntax error

Why?

[...very strange error messages skiped]

I will try to look into it, if I get an access to this compiler.
I was able to test with MSVC6.5, BCC command line, GCC3.2

Gennadiy.
___
Unsubscribe  other changes: http://lists.boost.org/mailman/listinfo.cgi/boost



Re: [boost] Regression test woes

2002-12-09 Thread David Abrahams
Rozental, Gennadiy [EMAIL PROTECTED] writes:

 #File: ..\libs\test\src\execution_monitor.cpp
 # ---
 # 157:  case _CRT_ASSERT:
 #   Error:   ^^^
 #   undefined identifier '_CRT_ASSERT'
 ### mwcc Compiler:
 # 161:  case _CRT_ERROR:
 #   Error:   ^^
 #   undefined identifier '_CRT_ERROR'

 What kind of C runtime hooks metrowerks supports?

I don't know. What's a C runtime hook?

 
 Errors caused tool to abort.
 
 call c:\tools\cwpro8.3\Other Metrowerks Tools\Command 
 Line Tools\cwenv.bat -quiet 

 Could I get these command line tools somewhere to install on my local PC for
 testing of Boost.Test?

You can buy one from Metrowerks, I guess.  If you ask nicely some
Booster may be able to get you an account where you have access to
them.

 #From: ..\libs\test\src\test_tools.cpp
 # 
 #  53:  voidshort_report( std::ostream where_to_,
 size_t indent_ = 0 );// short
 #   Error:
 ^^
 #   declaration syntax error

 Why?

Perhaps you didn't add the std:: qualification or you didn't include
stddef.h?

Those would be the obvious and usual reasons for a complaint about
size_t.

-- 
   David Abrahams
   [EMAIL PROTECTED] * http://www.boost-consulting.com
Boost support, enhancements, training, and commercial distribution

___
Unsubscribe  other changes: http://lists.boost.org/mailman/listinfo.cgi/boost