Re: JUnit sc_complex fails in Localized enviroment

Hi there,

On Sat, 2012-02-25 at 17:52 +0100, Maciej Rumianowski wrote:
> I hjave discovered a problem when checking libreoffice build (master),
> below is log. It seems that test is not fully localized.

        Interesting - what locale are you using out of interest ?

> Is "make check" meant to work in non-English enviroment?

        Yes, naturally :-)

> I had once such problem with filters_tests in Calc
> (http://nabble.documentfoundation.org/calc-filters-test-problem-in-Localized-enviroment-td3382023.html).
>  I don't know How Michael fixed it.

        That was fixed by:

commit 6f1dbfbe71b57f70fee88c49487436cb1e23ec6f
Author: Michael Meeks <michael.meeks@suse.com>
Date:   Fri Sep 30 14:57:15 2011 +0100

    set the core locale as well as the UI one to English

        But I think this problem is unrelated. Can you confirm that this passes
with a LANG=en_US.UTF8 set ? are you sure it's locale related ? :-)

> > There were 2 failures:
> > 1) checkEmptyCell(complex.cellRanges.CheckXCellRangesQuery)
> > java.lang.AssertionError:   Query column differences did not return the correct value.
> >     at complex.cellRanges.CheckXCellRangesQuery.checkEmptyCell(CheckXCellRangesQuery.java:179)
> > 2) checkFilledCell(complex.cellRanges.CheckXCellRangesQuery)
> > java.lang.AssertionError:   Query column differences did not return the correct value.
> >     at complex.cellRanges.CheckXCellRangesQuery.checkFilledCell(CheckXCellRangesQuery.java:207)

        Nasty. So - the JUnit stuff does a lot more heavy lifting, and (no
doubt) could more easily hit one of our multiple, duplicate ways of
fetching / acting on the translation / environment :-)

        'make check' and the JUnit stuff is Stephan's baby really.

        HTH,

                Michael.

-- 
michael.meeks@suse.com  <><, Pseudo Engineer, itinerant idiot