testsuite failures, no explanation
Adam Megacz
gcj@lists.megacz.com
Wed Aug 21 19:51:00 GMT 2002
Thanks for helping with this, David... hopefully soon I will have a
dedicated machine running 'make check' on cygwin every night, and
emailing people if they cause regressions.
- a
"Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com> writes:
> Wrong advice.
>> the patch you want will be similar to my patch of
>> 2002年05月06日 David.Billinghurst <David.Billinghurst@riotinto.com>
>> * testsuite/lib/libjava.exp (libjava_arguments): Don't link
> with -no-install on *-*-cygwin*.
>>>> -----Original Message-----
> From: Billinghurst, David (CRTS)
> Sent: Thursday, 22 August 2002 9:38
> To: Adam Megacz; java@gcc.gnu.org
> Subject: RE: testsuite failures, no explanation
>>> Dejagnu interprets any unexpected compiler output as a FAIL, so
> the the failure is (are) the lines
>> output is:
> libtool: link: warning: `-no-install' is ignored for i686-pc-mingw32
> libtool: link: warning: assuming `-no-fast-install' instead
>> These need to be pruned from the output by dejagnu. There are a number
> of places to do it, with different degrees of elegance.
>> Have a look at:
> libjava/testsuite/lib/libjava.exp
> /usr/share/dejagnu/target.exp
> and gcc/testsuite/lib/*.exp (and search for prune)
>> To get the libjava testsuite running natively on cygwin I need
> CVS dejagnu and had to make some crude hacks. I think I posted
> them to the list, but may have missed some.
>> -----Original Message-----
> From: Adam Megacz [mailto:gcj@lists.megacz.com]
> Sent: Thursday, 22 August 2002 8:56
> To: java@gcc.gnu.org
> Subject: testsuite failures, no explanation
>>>> Hey, I've finally got a cygwin-to-mingw cross up and running, and I
> can run 'make check'. Unfortunately, in libgcj.log, I get this:
>> Running /home/megacz/gcc/libjava/testsuite/libjava.compile/compile.exp ...
> Executing on host: /home/megacz/gcc-bin/i686-pc-mingw32/libjava/testsuite/../libtool --silent --tag=GCJ --mode=link /home/megacz/gcc-bin/gcc/gcj -B/home/megacz/gcc-bin/gcc/ --encoding=UTF-8 -B/home/megacz/gcc-bin/i686-pc-mingw32//libjava/ /home/megacz/gcc/libjava/testsuite/libjava.compile/ArrayClass.java -no-install --main=ArrayClass -g -L/home/megacz/gcc-bin/i686-pc-mingw32//libjava/.libs -L/home/megacz/gcc-bin/i686-pc-mingw32//boehm-gc/.libs -lm -o /home/megacz/gcc-bin/i686-pc-mingw32/libjava/testsuite/ArrayClass.exe (timeout = 300)
> spawn /home/megacz/gcc-bin/i686-pc-mingw32/libjava/testsuite/../libtool --silent --tag=GCJ --mode=link /home/megacz/gcc-bin/gcc/gcj -B/home/megacz/gcc-bin/gcc/ --encoding=UTF-8 -B/home/megacz/gcc-bin/i686-pc-mingw32//libjava/ /home/megacz/gcc/libjava/testsuite/libjava.compile/ArrayClass.java -no-install --main=ArrayClass -g -L/home/megacz/gcc-bin/i686-pc-mingw32//libjava/.libs -L/home/megacz/gcc-bin/i686-pc-mingw32//boehm-gc/.libs -lm -o /home/megacz/gcc-bin/i686-pc-mingw32/libjava/testsuite/ArrayClass.exe
> libtool: link: warning: `-no-install' is ignored for i686-pc-mingw32
> libtool: link: warning: assuming `-no-fast-install' instead
> output is:
> libtool: link: warning: `-no-install' is ignored for i686-pc-mingw32
> libtool: link: warning: assuming `-no-fast-install' instead
>> FAIL: ArrayClass compilation from source
> PASS: ArrayClass byte compilation
>>> The strange part is that the first command (which builds
> ArrayClass.exe) succeeds! Can anybody figure out why I get a FAIL in
> the log?
>> - a
>
--
"Experts arose from their own urgent need to exist." -- Murphy's laws
More information about the Java
mailing list