Boost logo

Boost :

Subject: Re: [boost] [regression] Regression tests for 'develop' has error
From: Edward Diener (eldiener_at_[hidden])
Date: 2016-01-12 15:06:30


On 1/11/2016 2:04 PM, Louis Dionne wrote:
> Jürgen Hunold <jhunold <at> gmx.eu> writes:
>
>>
>> Hi Rene,
>>
>> Am Sunday 10 January 2016, 20:43:38 schrieb Rene Rivera:
>>>> Every time I try and replicate this it works (and temporarily fixes the
>>>> problem as a side effect). So it's been, so far, impossible to determine
>>>> the cause of the problems.
>>>
>>> PS. Although, one time it was caused by ultra large results (from long
>>> logs), IIRC
>>
>> This might be coincidence, but AFAIK the test matrix went out of business
>> shortly after Hana (and then Dll) were added. And I'm seeing quite a lot of
>> (unexpected) errors from Hana here.
>
> Hana has a huge number of tests, and the output could be gigantic when
> compile-time errors happen due to compiler spew. Hana is not supposed to
> compile on most compilers in the test matrix; would it be possible to
> somehow mark failures on these compilers as expected, or to skip running
> the tests altogether? Would that help in any way?

You can definitely skip running tests in your jam file for all the
compilers which cannot handle Hana. You can use predef-require to do
this or you can use the Boost build 'alias' feature.


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk