build failure icu-54.1-2.fc23 armv7hl

[Eike Rathke]

Hi Peter,

On Monday, 2015-04-27 11:27:07 +0100, Peter Robinson wrote:

> > icu-54.1-2.fc23 armv7hl had a build failure when executing test cases
> > for number formatting, see
> > http://koji.fedoraproject.org/koji/taskinfo?taskID=9469050
> > Same happened building icu-54.1-3.fc22, but the logs are gone.
> >
> > However, I tried to reproduce building armv7hl f22, but encountered no
> > error whatsoever.
> 
> How did you try and reproduce it? Another build in koji, used mock or
> some other means?

I used a Red Hat internal machine, building the f22 src on f21. So maybe
the difference is that one has to build on f22 or rawhide to reproduce
the error.

> koji doesn't have outbound network connectivity so
> if a test relies on being able to connect to an outside resource or
> similar it might fail, but I can't interpret the output of the failure
> above to work out what it was trying to do.

There should be no network connectivity involved. The failure was in
scientific number formatting such that a number is formatted with one
order of magnitude off, though the actual number is correct, for
example:

ERROR: Expected -10.000E-01 Got: -1.0000E00
ERROR: Expected 12.346E07 Got: 1.2346E08

The tests are in icu/source/test/intltest/numfmtst.cpp
NumberFormatTest::TestExponential() and
NumberFormatTest::TestScientific()
Only some of the tests fail though..


  Eike

-- 
LibreOffice Calc developer. Number formatter stricken i18n transpositionizer.
GPG key "ID" 0x65632D3A - 2265 D7F3 A7B0 95CC 3918  630B 6A6C D5B7 6563 2D3A
Better use 64-bit 0x6A6CD5B765632D3A here is why: https://evil32.com/
Care about Free Software, support the FSFE https://fsfe.org/support/?erack
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.fedoraproject.org/pipermail/devel/attachments/20150428/174d38e6/attachment.sig>