Re: [boost] [Interprocess] Lengthy test runtimes

From: Boris Gubenko (Boris.Gubenko_at_[hidden])
Date: 2007-08-07 17:52:26

FWIW, three interproces library tests hang on HP-UX/ia64:

        . barrier_test
        . condition_test
        . upgradable_mutex_test

I did not have a chance to investigate it and, probably, won't have for
a while. When I start regression tests, I also start a script which waits
for these tests and kills them ('/usr/sbin/fuser -ku test_executable').
And no, I don't run tests for multiple toolsets.

Thanks,
  Boris

----- Original Message ----- 
From: "Rene Rivera" <grafikrobot_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Tuesday, August 07, 2007 4:50 PM
Subject: Re: [boost] [Boost][Interprocess] Lengthy test runtimes

Ion Gaztañaga wrote:
> K. Noel Belcourt wrote:
>> Because I run the boost regression tests like this
>>
>> python regression.py --runner="Sandia-gcc" --mail=kbelco_at_[hidden] -- 
>> bjam-toolset=gcc --pjl-toolset=gcc --toolsets="gcc-3.4.3,gcc-4.0.1" -- 
>> bjam-options=-j4
>>
>> So bjam is clearly building / running both named_recursive tests at
>> the same time.  Am I not permitted to run the boost regression tests
>> for multiple toolsets in a single invocation?
>
> Ummm. Both test might be writing to the same shared memory overwriting
> their contents. Can you try to run them separately? Otherwise I might
> need to create a unique shared memory name (I create a new name for each
> compiler name, but not for different versions of the same compiler).

Ion, you'll have to fix the names. We went through a fair amount of work
to make it possible to run test in parallel. It's an important feature
as it reduces testing time considerably, even if you only have 1 CPU.

Note, it may not only be for different compiler versions. If you have
different tests that can clobber each other, it's possible they will
clobber each other when run on a single compiler test run is they tests
run at the same time.

-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com
-- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost