|
|
Boost Testing : |
From: Misha Bergal (mbergal_at_[hidden])
Date: 2006-09-18 00:12:49
David Abrahams wrote:
> Misha Bergal <mbergal_at_[hidden]> writes:
>
>
>> David Abrahams <dave <at> boost-consulting.com> writes:
>>
>>
>>> I checked in fixes for libs/parameter/test/maybe.cpp on 2006/09/13 03:15:16
>>> Yet the test results for vc-6.5 generated almost 7 hours later, at
>>> 2006-09-13 10:00:15, don't reflect the changes.
>>>
>>>
>>>
>> http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/output/metacomm-bin-boost-libs-parameter-test-maybe-test-vc-6_5-debug-threading-multi_release.html
>>
>> Looks OK now.
>>
>> The problem is that metacomm cycle now takes much more more than usual. It seems
>> that Digital Mars compiler hangs compiling certain tests, increasing the cycle
>> time to up to 12 hours. Until I localize the problem, expect about 15 hours for
>> metacomm results to propagate to report.
>>
>
> I suggest removing Digital Mars from that cycle until you get it
> worked out. It doesn't make sense to penalize all people waiting on
> test results just for that one compiler. You could just run a separate
> test process for DM, for example
Sorry, after closer examination I can see that the cycle takes as long
as it used to - about 4 hours for metacomm2 (dmc), and another 13 for
metacomm ( borland-5_6_4, borland-5_8_2, vc-6_5, vc-6_5-stlport, vc-7_0
and cw-8_3).
-- Misha Bergal MetaCommunications Engineering