harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mark Hindess <mark.hind...@googlemail.com>
Subject Re: [classlib][portlib] the portlib tests failed to compile on Windows platform
Date Mon, 19 Feb 2007 19:40:51 GMT

On 19 February 2007 at 22:16, "Alexei Zakharov"
<alexei.zakharov@gmail.com> wrote:
> Hi guys,
> 
> Excuse me, I can be off-track here. Just trying to run portlib tests
> on Windows for the first time.

Thanks.  As I mentioned before, I don't have a windows test machine at
the moment so I was hoping someone would take a look at this.  I've
checked in a fix (r509310) that I hope will get things a little further.

-Mark.

> I've run:
> 
> ant -Dbuild.module=portlib -Dtest.portlib=true test
> 
> And got:
> ---
> test-modules:
> 
> -compile-native-tests:
>      [exec]     link /NOLOGO  /INCREMENTAL:NO /NOLOGO -subsystem:console,4.0 
> -su
> bsystem:console /debug /opt:icf /opt:ref   -out:../init.exe -machine:i386 set
> arg
> v.obj   ..\shared\init.obj  C:\Java\harmony\enhanced\classlib\trunk\deploy\li
> b\h
> ythr.lib C:\Java\harmony\enhanced\classlib\trunk\deploy\lib\hycommon.lib C:\J
> ava
> \harmony\enhanced\classlib\trunk\deploy\lib\hysig.lib kernel32.lib  wsock32.l
> ib
> advapi32.lib
>      [exec] init.obj : error LNK2019: unresolved external symbol _hyport_init
> _li
> brary referenced in function _main
>      [exec] ../init.exe : fatal error LNK1120: 1 unresolved externals
>      [exec] NMAKE : fatal error U1077: 'link' : return code '0x460'
>      [exec] Stop.
> 
> BUILD FAILED
> C:\Java\harmony\enhanced\classlib\trunk\build.xml:167: The following error oc
> cur
> red while executing this line:
> C:\Java\harmony\enhanced\classlib\trunk\make\build-test.xml:52: The following
>  er
> ror occurred while executing this line:
> C:\Java\harmony\enhanced\classlib\trunk\make\properties.xml:341: The followin
> g e
> rror occurred while executing this line:
> C:\Java\harmony\enhanced\classlib\trunk\modules\portlib\build.xml:148: The fo
> llo
> wing error occurred while executing this line:
> C:\Java\harmony\enhanced\classlib\trunk\make\properties.xml:356: exec returne
> d:
> 2
> ---
> 
> I believe something is wrong with my config. I'll be happy if someone
> could help.
> 
> Thanks,
> 
> 2007/1/24, Mark Hindess <mark.hindess@googlemail.com>:
> >
> > On 24 January 2007 at 19:48, "Vladimir Ivanov" <ivavladimir@gmail.com> wrot
> e:
> > >
> > > On 1/24/07, Mark Hindess <mark.hindess@googlemail.com> wrote:
> > > >
> > > > On 24 January 2007 at 18:54, "Vladimir Ivanov" <ivavladimir@gmail.com>
> > > > wrote:
> > > > >
> > > > > Seems, it does not work :(
> > > >
> > > > What is the output in the generated test report?  You can browse to it
> > > > but it will probably be in a file called:
> > > >
> > > > build/test_report/html/native/portlib/1_hyfile-out.txt
> > > >
> > > > FYI: I've added more verbose errors to the latest version in svn.
> > >
> > > Output is (which file should be deleted?):
> > > hyfile:
> > >   portlib initialized
> > >   fd = 4012
> > >   hyfile_write wrote 7 bytes
> > > hyfile_move failed (183) Cannot create a file when that file already exis
> ts.
> >
> > In general, modules/portlib/*tmp*.
> >
> > Oddly enough the next item on my todo list (stack) was to look at
> > adding a cleanup step to remove any created files when the tests fail.
> >
> > This seems to be one of those jobs where my todo list just grows the
> > more I look in to it.  First I spot that when move_file fails on Linux
> > the portlib hyerror message is not set.  Then when looking at that I see
> > a pointless copy in errorMessage caused by using the non-thread-safe
> > strerror call instead of strerror_r.  Then I decide to confirm the most
> > portable way to use strerror_r by testing on Linux, Aix and FreeBSD...
> >
> > Sorry for the breakages (hopefully fixed now - except the
> > check-for-empty one which shouldn't occur in normal use?) and thanks
> > very much for letting me know what was failing on windows.
> >
> > -Mark.
> >
> > > > Output is (note, it does not depend on if/unless in the string <target
> > > > > name="-test-module" unless/if="test.portlib"):
> > > > >
> > > > > support-jar:
> > > > >       [jar] Building jar:
> > > > > C:\harmony.top\classlib\trunk\deploy\build\test\support.jar
> > > > > test-modules:
> > > > > -compile-native-tests:
> > > > > -run-native-tests:
> > > > >     [mkdir] Created dir: C:\harmony.top\classlib\trunk\build\test_rep
> ort
> > > > >      [echo] init: passed
> > > > >      [echo] hyerror: passed
> > > > >      [echo] hytime: passed
> > > > >      [exec] Result: 1
> > > > >      [echo] hyfile: FAILED
> > > > > touch-errors-file:
> > > > > -test-module:
> > > > > -check-for-empty:
> > > > > -process-empty:
> > > > > process-empty:
> > > > > full-report:
> > > > >
> > > > > but if you add 'if="test.portlib"' to the targets
> > > > '-compile-native-tests'
> > > > > and '-run-native-tests' no tests will be run and
> > > >
> > > > Hmm... I didn't realise depends were executed even when if/unless
> > > > conditions fail.  Fixed now.
> > >
> > >
> > > Thanks, the "ant -Dbuild.module="p*" test" passed for me now.
> > >
> > > > build failed on the
> > > > > '-check-for-empty' target only with message "build\test_report not
> > > > found".
> > > >
> > > > This is really a bug/feature of make/build-test.xml.
> > > >
> > > > Regards,
> > > > Mark.
> > > >
> > > > > On 1/24/07, Mark Hindess <mark.hindess@googlemail.com> wrote:
> > > > > >
> > > > > >
> > > > > > On 24 January 2007 at 15:53, "Vladimir Ivanov" <ivavladimir@gmail.c
> om>
> > > > > > wrote:
> > > > > > >
> > > > > > > Thanks Mark,
> > > > > > > now these tests compiled on windows platform but one of
them fail
> ed:
> > > > > > > -run-native-tests:
> > > > > > >     [mkdir] Created dir:
> > > > C:\harmony.top\classlib\trunk\build\test_report
> > > > > > >      [echo] init: passed
> > > > > > >      [echo] hyerror: passed
> > > > > > >      [echo] hytime: passed
> > > > > > >      [exec] Result: 1
> > > > > > >      [echo] hyfile: FAILED
> > > > > > >
> > > > > > > Also these tests continue to hang up on x86_64 platform.
> > > > > > > May be special key (-Drun.native.tests=true) will be useful
for t
> he
> > > > time
> > > > > > of
> > > > > > > development infrustructure for first native tests?
> > > > > > >  thanks, Vladimir
> > > > > >
> > > > > > I've added test.portlib but I'd really like to understand why
it fa
> ils
> > > > > > on windows.  Please can someone test it with:
> > > > > >
> > > > > > ant -Dbuild.module=portlib -Dtest.portlib=true test
> > > > > >
> > > > > > and then look at:
> > > > > >
> > > > > > build/test_report/html/native/portlib/1_hyfile-out.txt
> > > > > >
> > > > > > Regards,
> > > > > > -Mark.
> > > >
> > > >
> > > >
> > >
> > > ------=_Part_3012_2817132.1169646539227--
> 
> 
> -- 
> Alexei Zakharov,
> Intel ESSD
> 



Mime
View raw message