Fwd: Errors in test matrix

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Fwd: Errors in test matrix

Boost - Dev mailing list
[I haven't received a reply on the boost-testing list, so I'm resending
to the general list in the hopes that someone can help.]



-------- Forwarded Message --------
Subject: Errors in test matrix
Date: Wed, 31 Oct 2018 09:41:05 -0400
From: Stefan Seefeld <[hidden email]>
To: Running Boost regression tests <[hidden email]>



Hello,

[I'm not entirely sure this is the right place for the report. Please
let me know if it isn't, and where to post instead.]

On
https://www.boost.org/development/tests/develop/developer/numeric-ublas.html,
I'm looking at failures in the "Sandia" columns. Following one link
takes me to

https://www.boost.org/development/tests/develop/developer/output/Sandia-gcc-5-4-0-c++11-boost-bin-v2-libs-numeric-ublas-test-begin_end-test-gcc-4-4-7-debug-visibility-hidden.html,
which demonstrates a couple of issues:

* What compiler version is this using ? The URL contains both "5.4.0" as
well as "4.4.7".

* The actual error reads "unrecognized command line option -std=c++11".
Given that Boost.uBLAS itself doesn't specify this flag, I assume it's
part of the test runner's configurations. However, it seems to be used
with a compiler version that doesn't support it (though again, it isn't
clear which compiler version we are actually talking about).


Unfortunately there are quite a large number of columns with similarly
generic errors, i.e. errors more likely to be caused by invalid set ups,
rather than genuine errors in the library (libraries, as I'm seeing
similar errors in other libraries).

What can I do about this ? Ignore the errors ? It would be good if the
metainfo that is displayed with the test runners included a point of
contact (mail address ?) to report such errors.

Thanks,

Stefan

--

       ...ich hab' noch einen Koffer in Berlin...
     


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Reply | Threaded
Open this post in threaded view
|

Re: Fwd: Errors in test matrix

Boost - Dev mailing list
On Sun, Nov 4, 2018 at 12:16 AM Stefan Seefeld via Boost
<[hidden email]> wrote:
>
> On
> https://www.boost.org/development/tests/develop/developer/numeric-ublas.html,
> I'm looking at failures in the "Sandia" columns. Following one link
> takes me to
>
> https://www.boost.org/development/tests/develop/developer/output/Sandia-gcc-5-4-0-c++11-boost-bin-v2-libs-numeric-ublas-test-begin_end-test-gcc-4-4-7-debug-visibility-hidden.html,
> which demonstrates a couple of issues:

I don't see any gcc columns for Sandia in the current table and the
link to the test output page doesn't work for me. Maybe you were
looking at some stale data.

> * What compiler version is this using ? The URL contains both "5.4.0" as
> well as "4.4.7".

The compiler version typically comes from the tester name at the
header/footer of the table. If you click on it, you will see
additional info regarding the tester, although specifically for Sandia
there is not much useful there.

You can see some more info about the compiler by viewing the output of
the config_info test of Boost.Config:

https://www.boost.org/development/tests/develop/developer/config.html

> * The actual error reads "unrecognized command line option -std=c++11".
> Given that Boost.uBLAS itself doesn't specify this flag, I assume it's
> part of the test runner's configurations. However, it seems to be used
> with a compiler version that doesn't support it (though again, it isn't
> clear which compiler version we are actually talking about).

The flag is most likely added by the tester and if so, that is a
tester configuration error.

> What can I do about this ? Ignore the errors ? It would be good if the
> metainfo that is displayed with the test runners included a point of
> contact (mail address ?) to report such errors.

Normally, the contact info is included in the tester info page that I
mentioned earlier. Unfortunately, this is not the case for Sandia. I
think, those testers are run by Noel Belcourt (kbelco at sandia.gov).
He sometimes appears on this list.

Given that currently the faulty columns are not present, the problem
might have been fixed already (as it was reported earlier on the
list), and the columns were removed.

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost