On 2015/09/20 13:45, Peter Kane wrote: > Hello > > I wanted to set up some Japanese input so I tried a few things out on > an i386 test machine running -current where they worked (more or less) > OK. When I tried to migrate some of them across to an amd64 machine > running -current I got segfaults instead. > > Jless segfaults immediately, while jvim runs OK until I select romaji > input mode. Jserver seems to be running OK and I have > ja-onew-wnn4-canna installed. > > Here is some gdb output if that is of any use. Both are running in > kterm (where cat outputs sample kana text OK). > > jless > Starting program: /usr/local/bin/jless > (no debugging symbols found) > [INS] > Program received signal SIGSEGV, Segmentation fault. > 0x000001db1b51f082 in winch () from /usr/local/bin/jless > > > ja-jvim-2.0rp2-wnn-canna > OK until I start romaji input mode (eg C-space): > > Program received signal SIGSEGV, Segmentation fault. > _citrus_none_ctype_mbrtowc (pwc=0x7f7ffffdf158, > s=0xffffffffb08a03b2 <Address 0xffffffffb08a03b2 out of bounds>, n=1, > pspriv=0x7f7ffffdf080) at /usr/src/lib/libc/citrus/citrus_none.c:73 > 73 /usr/src/lib/libc/citrus/citrus_none.c: No such file or directory. > ~ in /usr/src/lib/libc/citrus/citrus_none.c
Looking at compiler warnings while building Wnn, it seems unlikely that it will work on 64-bit machines.
