The Trunk: KernelTests-pre.310.mcz

Eliot Miranda-2 wrote

So shouldn't the code test for being on the Windows platform and not answer
it as an expected failure on linux and Mac OS X?

On Mon, Aug 15, 2016 at 4:23 PM, Nicolas Cellier <
[hidden email]> wrote:

> It only fails for windows VM if using msvcrt.dll
> because Microsoft math library is bugged...
>
> 2016-08-16 1:07 GMT+02:00 Levente Uzonyi <[hidden email]>:
>
>> On Mon, 15 Aug 2016, [hidden email] wrote:
>>
>> Patrick Rein uploaded a new version of KernelTests to project The Trunk:
>>> http://source.squeak.org/trunk/KernelTests-pre.310.mcz
>>>
>>> ==================== Summary ====================
>>>
>>> Name: KernelTests-pre.310
>>> Author: pre
>>> Time: 15 August 2016, 11:25:58.171656 pm
>>> UUID: 38fde71e-1444-6b4f-9d3b-24221c24a5d2
>>> Ancestors: KernelTests-ul.309
>>>
>>> Recorded the wrong float underflow as an expected failure as it is going
>>> to be the expected behavior (somewhat) in the upcoming release
>>>
>>
>> I've never seen that test failing. Is it a windows specific failure?
>>
>> Levente
>>
>>
>>
>>> =============== Diff against KernelTests-ul.309 ===============
>>>
>>> Item was added:
>>> + ----- Method: FloatTest>>expectedFailures (in category
>>> 'characterization') -----
>>> + expectedFailures
>>> +
>>> +       ^ #(testTimesTwoPowerGradualUnderflow) !
>>>
>>>
>>>
>>>
>>
>
>
>
>


--
_,,,^..^,,,_
best, Eliot

On Tue, Aug 16, 2016 at 11:16 PM Nicolas Cellier <[hidden email]> wrote:

BTW, VM produced by appveyor automaton doesn't suffer from the problem, at least since we refactored the makefile and removed explicit msvcrt library linking.

So whether compiled with gcc or clang, the win32 failure will disappear as soon as we declare a newer vm as stable.

Does this mean we can/should use a recent win32 built by AppVeyor for the upcoming Squeak release?

 

2016-08-16 18:43 GMT+02:00 marcel.taeumel <[hidden email]>:

Eliot Miranda-2 wrote
> So shouldn't the code test for being on the Windows platform and not
> answer
> it as an expected failure on linux and Mac OS X?
>
> On Mon, Aug 15, 2016 at 4:23 PM, Nicolas Cellier <

> nicolas.cellier.aka.nice@

>> wrote:
>
>> It only fails for windows VM if using msvcrt.dll
>> because Microsoft math library is bugged...
>>
>> 2016-08-16 1:07 GMT+02:00 Levente Uzonyi &lt;

> leves@.elte

> &gt;:
>>
>>> On Mon, 15 Aug 2016,

> commits@.squeak

>  wrote:
>>>
>>> Patrick Rein uploaded a new version of KernelTests to project The Trunk:
>>>> http://source.squeak.org/trunk/KernelTests-pre.310.mcz
>>>>
>>>> ==================== Summary ====================
>>>>
>>>> Name: KernelTests-pre.310
>>>> Author: pre
>>>> Time: 15 August 2016, 11:25:58.171656 pm
>>>> UUID: 38fde71e-1444-6b4f-9d3b-24221c24a5d2
>>>> Ancestors: KernelTests-ul.309
>>>>
>>>> Recorded the wrong float underflow as an expected failure as it is
>>>> going
>>>> to be the expected behavior (somewhat) in the upcoming release
>>>>
>>>
>>> I've never seen that test failing. Is it a windows specific failure?
>>>
>>> Levente
>>>
>>>
>>>
>>>> =============== Diff against KernelTests-ul.309 ===============
>>>>
>>>> Item was added:
>>>> + ----- Method: FloatTest>>expectedFailures (in category
>>>> 'characterization') -----
>>>> + expectedFailures
>>>> +
>>>> +       ^ #(testTimesTwoPowerGradualUnderflow) !
>>>>
>>>>
>>>>
>>>>
>>>
>>
>>
>>
>>
>
>
> --
> _,,,^..^,,,_
> best, Eliot

Hi Eliot,

yes, I already did that.

Best,
Marcel



--
View this message in context: http://forum.world.st/The-Trunk-KernelTests-pre-310-mcz-tp4911102p4911371.html
Sent from the Squeak - Dev mailing list archive at Nabble.com.