|
Boost Testing : |
Subject: Re: [Boost-testing] Result masking behaviour for libraries marked"unusable"
From: Armand Li (armandli_at_[hidden])
Date: 2013-10-02 13:41:15
I spotted a error in my patch right after sending it out ( should return
percentage failures not success for unussables ). here is the updated
patch.
(See attached file: boost-tool-patch2.patch)
Armand
From: Hubert Tong/Toronto/IBM
To: Running Boost regression tests <boost-testing_at_[hidden]>,
Cc: "Boost-Testing" <boost-testing-bounces_at_[hidden]>, Armand Li/Toronto/IBM_at_IBMCA
Date: 2013/10/02 10:12 AM
Subject: Re: [Boost-testing] Result masking behaviour for libraries marked"unusable"
Hi all,
We have a locally tested patch ready which shows the percentage passing for
libraries which are marked unusable and also the percentage failing for
other libraries.
My colleague will be sending it shortly.
-- Hubert Tong
Hubert S. K. Tong
IBM Canada Ltd.
8200 Warden Avenue
Markham, Ontario L6G 1C7
Canada
E-mail: hstong_at_[hidden]
From: "John Maddock" <john_at_[hidden]>
To: "Running Boost regression tests"
<boost-testing_at_[hidden]>,
Date: 22-08-2013 11:51 AM
Subject: Re: [Boost-testing] Result masking behaviour for libraries
marked"unusable"
Sent by: "Boost-Testing" <boost-testing-bounces_at_[hidden]>
> I am running the AIX tester and have upgraded to version 12.1.0.4 of the
> IBM compiler.
> I believe that the current treatment of libraries marked unusable,
> especially when the tests are mostly passing, is unhelpful both for
myself
> and to the community.
> Since the results are masked out of the status page, how are people
> supposed to gauge for themselves whether or not the library is usable for
> them?
>
> For example, the geometry/extensions tests all pass on the test results
> for
> the AIX runner except for issues with buffer_inserter.hpp.
> If a user wanted to use geometry/extensions and did not have a need for
> the
> functionality tied to buffer_inserter.hpp, they are currently impeded
from
> finding the actual status.
I think it's a good point - as library authors we have no way of knowing
when a compiler will jump from "unsupportable" to "compliant" so showing
the
actual results would make sense.
John.
_______________________________________________
Boost-Testing mailing list
Boost-Testing_at_[hidden]
http://lists.boost.org/mailman/listinfo.cgi/boost-testing