On Thu, Oct 18, 2007 at 05:28:21PM +0100, Francis Tyers wrote: > > It would seem that while this fixes the problem on amd64, it causes a > new problem on ia32. See the bug page here: > > http://xixona.dlsi.ua.es/cgi-bin/bugzilla/show_bug.cgi?id=26 > > For more info. Any more thoughts ?
I've tried to reproduce the problem with my patch on i386, and it worked without problems for me. I've tried it both with the en-ca en es-ca. The only thing valgrind can come up with is: ==10432== Conditional jump or move depends on uninitialised value(s) ==10432== at 0x446F758: TRXReader::procDefCats() (trx_reader.cc:384) ==10432== by 0x4471C08: TRXReader::read(std::string const&) (trx_reader.cc:177) ==10432== by 0x8048C34: (within /usr/bin/apertium-preprocess-transfer) ==10432== by 0x627C04F: (below main) (in /lib/libc-2.6.1.so) Which I seem to have on amd64 too anyway. I think "type" is not initialised, I only see an assignment in step(), and step() is called after procDefCats(). I only see step() being called after type is checked. Kurt -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]