[oleaut32/tests] Error when running tests on Wine, anyone?

[Paul Vriens]

Paul Vriens wrote:
> Detlef Riekenberg wrote:
>> On So, 2007-08-12 at 10:50 +0200, Stefan Leichter wrote:
>>>>> Can somebody confirm if all oleaut32 tests run correctly on Wine. 
>>>> detlef at p4:~/wine.cvs/bin$ wine dlls/oleaut32/tests/oleaut32_test.exe.so
>>>> vartest
>> ...
>>>> vartest.c:3129: Test failed: got VT_R8(-145) expected VT_R8(-1.4)
>>>> vartest: 1031878 tests executed (0 marked as todo, 1 failure), 0
>>>> skipped.
>>>> Intel P4 2GHz / Ubuntu 6.06
>>> Should we think about localisation? 
>>
>> That's the reason for the failure!
>> LANG="en_GB.UTF-8" wine ... => 0 failures
>> Also for "en_GB", "it_IT", "fr_FR", "gr_GR", ....
>>
>>> Of course we are running both debian base distributions. This might 
>>> be another item to look on.
>>
>> Not needed.
>> Our tests are not correct for w2k and below.
>> My w2k_sp4(german) has 539 failures and includes the failure from wine:
>>> vartest.c:3129: Test failed: got VT_R8(-145) expected VT_R8(-1.4)
>>
>>
>> The WRT-Resultpage has 536 failures for w2k:
>> http://test.winehq.org/data/200708071000/2000_W2KProf-Admin/oleaut32:vartest.txt 
>>
>>
>>
>>
> OK, we have a few results now and it we know that localization is 
> important. This however doesn't fix it for me, so there are more variables.
> 
> I'm running on FC7 x86_64 with gcc 4.1.2. If people can post there 
> versions etc.. we can maybe come up with the other variables (this also 
> means people where the tests succeeds).
> 
> So far:
> 
> - localization influences the tests (which make sense because of the '.' 
> in 1.45 which is a ',' in other locales)
> - I have 2 failures reported (with the same results) on a FC6/FC7 
> x86_64. So maybe the 64-bit is a variable as well.
> 
> Cheers,
> 
> Paul.
> 
Just installed a new F7 (x86) and the failures are present there as well. So 
it's most likely not a 64bit thingy. Could thus be gcc related?

Could we rewrite the test:

VARROUND(DATE,-1.45,1,DATE,-1.4);

as :
VARROUND(DATE,(-145/100),1,DATE,-1.4);

to overcome the decimal separator issue. Or do we need other means to construct 
the 1.45/1,45?

Cheers,

Paul.