[Boost-testing] Re: Re: Completely out of date results...

From: Paul Baxter (pauljbaxter_at_[hidden])
Date: 2005-05-03 03:36:57

Jeff Garland wrote:
> On Mon, 2 May 2005 00:41:03 +0100, Paul Baxter wrote
>> I cannot run the tests very often or regularly. I wouldn't consider
>> myself a 'regression tester' though I am interested in the state of
>> Boost on Windows and Linux using icc and had hoped to give a little
>> back when possible.
>>
> Given that sslapeta is running intel-win32-8_1, I'd be interested in 
> knowing
> if you have the same versions or different.  Maybe if you are going to run 
> at
> a low frequency you could make sure to do something unique like running in
> release mode instead of debug.  This would add another dimension even 
> though
> it is low frequency that otherwise wouldn't get covered.

Just updated from icc v8.1.025 to latest 8.1.027 for win32
Configured as a release regression test

Results:
filter_iterator_test core dumps - send to Microsoft etc.
random_test has to be manually halted (or takes > 1hr to run on a 3.2 P4)

If I can help others in diagnosing this, let me know.

PS On previous debug test I was still getting the memory leaks as I 
previously reported. Were there changes to the string libs and some mods to 
threading and its relation to boost.test (other than the speed issue) that 
should have improved the situation.