Am Samstag, 14. Oktober 2006 20:44 schrieb chromatic: > Hi there, > > Somewhere in the past three checkins is what appears to be an endianness > bug; with them, many many tests fail:
Indeed - sorry, fixed r14926. I've leanred: on BE boxen you can't format a 'long long' with '%ld' obviously, which is breaking a lot of 'print I0' and such usage. Parrot_sprintf_format is now using variations of '%lld', which OTOH might break non-susv3 compliant compilers, which don't support the 'll' modifier. If this is a problem, we could a) revert major parts of r14924 & r14926 and implement all the nastiness of e.g. $ perl -e'printf("!%3.2d!\n", 1)' ! 01! in spf_render.c, which I wanted to avoid. or b) test for a proper formatting char for HUGEINTVAL. Thanks for bug reporting, leo