Re: [boost] Boost regression notification (2006-11-05 [RC_1_34_0])

From: Anthony Williams (anthony_w.geo_at_[hidden])
Date: 2006-11-05 18:47:30

"Peter Dimov" <pdimov_at_[hidden]> writes:

> Anthony Williams wrote:
>
>> Do failures like this one not warrant a mention? (bind on vc 7.1)
>>
>> http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/output/JustSoftwareSolutions-bin-boost-libs-bind-test-mem_fn_eq_test-test-vc-7_1-release-threading-multi.html
>
> This failure is caused by VC's linker folding duplicate identical functions 
> into one (which is why the test only fails in release mode). I don't know 
> what can be done about it. It's probably not a problem in practice. 

Maybe it should be marked up as "expected failure" or something,
then. Currently the RC-1.34 summary at
http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/summary.html
shows it as a red "Broken" box, yet it's not on the regression notification
report.

There's a few others like that, too.

http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/output/JustSoftwareSolutions-bin-boost-libs-spirit-test-symbols_add_null-test-vc-7_1-release-threading-multi.html
http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/output/JustSoftwareSolutions-bin-boost-libs-numeric-interval-test-test_float-test-vc-8_0-release-threading-multi.html
http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/output/JustSoftwareSolutions-bin-boost-libs-iterator-test-filter_iterator_test-test-vc-8_0-release-threading-multi.html
http://engineering.meta-comm.com/boost-regression/CVS-RC_1_34_0/developer/output/JustSoftwareSolutions-bin-boost-libs-iterator-test-lvalue_concept_fail-test-vc-7_1-release-threading-multi.html

All these are from the tests I ran yesterday; I haven't cross-checked to see
if there are similar cases from other people's test runs.

The important thing from my point of view is that the regression report
doesn't replicate the summary page. This means that people won't be
investigating potentially broken things.

Secondly, the results from my regression runs don't seem to tally with those
from RudbekAssociates-V2, despite my having a reasonably standard installation
of MSVC7.1 and MSVC8.0Express on Windows 2000. What's up with that.

Anthony
-- 
Anthony Williams
Software Developer
Just Software Solutions Ltd
http://www.justsoftwaresolutions.co.uk