[Libreoffice] make check fails in ScFiltersTest

[Markus Mohrhard]

Hello Regina,

2011/10/20 Regina Henschel <rb.henschel at t-online.de>:
> Hi Markus,
>
> Markus Mohrhard schrieb:
>>
>> Hello Regina,
>>
>> 2011/10/20 Regina Henschel<rb.henschel at t-online.de>:
>>>
>>> Hello Markus,
>>>
>>> Markus Mohrhard schrieb:
>>>>
>>>> Hello Regina,
>>>>
>>>> could you apply these two patches, rebuild sc and post the log file:
>>>> workdir/unxlngx6.pro/CppunitTest/sc_filters_test.test.log where
>>>> unxlngx6.pro is the corresponding windows folder.
>>>>
>>>>
>>>>
>>>> http://cgit.freedesktop.org/libreoffice/core/commit/?id=cda653d7d9038fdcab72cfcae37d3d20463693d5
>>>>
>>>>
>>>> http://cgit.freedesktop.org/libreoffice/core/commit/?id=d9604fa1b4123da145fc9b52901b42888d25ca3b
>>>>
>>>> I hope that these may give me more hints why these test fail at a
>>>> windows build but not at a linux or mac build.
>>>>
>>>> Thanks for your error log, it shows that at least our language
>>>> settings are correct.
>>>
>>> The log-file is attached.
>>
>> Thanks a lot, that helps me a lot. Could you just check if Sheet1.A2
>> in sc/qa/unit/data/xls/formats.xls contains 40.10% or 0.00% if you
>> open the file with your build. I still don't understand why the
>> filters-test import goes wrong in this cell.
>
> The content is 3,23635881614008E-229%.
> The display is 0,00%.
> (It is comma because my local is German.)
>
> Is the source of the values a file too?
>
I don't understand what you mean with this question.

All in all, thanks a lot for your help. I think you helped to find
another calc bug that only shows in a windows build.

Kohei, Eike: Any ideas why our xls import of these cells goes wrong in
a windows build. It seems that none of our Linux or mac builds has a
problem with this cell import. I'm looking if perhaps my mingw build
shows the same but otherwise I have no idea. The content looks a bit
like the fact that we added a non initialized double variable.

Markus