On Thu, Jan 17, 2008 at 05:54:06PM -0500, Jack Lloyd wrote: > On Thu, Jan 17, 2008 at 11:42:53PM +0100, Pavel Cahyna wrote: > > > It compiles; ./check output is attached. I interrupted it after 7 minutes > > when it did not seem to make any progress. > > If division/modulo is broken as badly as the tests suggest, it's > highly likely that one of the power_mod algorithms had gone into an > infinte loop. In particular it looks like it is possible with this bug > for a % n > n -- that is going to cause problems. > > The most likely files to be miscompiled that would break things in > this way are mp_misc.cpp, divide.cpp, then the other mp_* files, then > the big_* files (in order of likelyhood).
I compiled mp_misc.cpp with -O0, relinked with the rest and all test now pass. Have a look at http://colabti.de/irclogger/irclogger_log/monotone?date=2007-06-15,Fri#l165 too, there is a talk about other compilers mis-optimising botan. Pavel _______________________________________________ Monotone-devel mailing list Monotone-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/monotone-devel