On Tue, Jan 23, 2001 at 09:02:52AM -0600, Jarkko Hietaniemi wrote: > Very possible. But how about them op/numconvert and op/int failures, eh? :-) you didn't say anything on Sunday about carrying on being my guilty conscience indefinitely. w.r.t. op/int "don't shoot the messenger" - I only wrote a new test. Abigail kindly built release 5.7.0 and it fails there too. So it's nothing to do with changes I made. It looks like it's the C library or the compiler. I'm stumped as to what exactly it might be, or how to detect it. Likewise Irix on op/inc still doesn't have a proven cause, or numconvert on x86 Solaris (for 64 bit integers, IIRC) I don't know what to do about any of these, apart from write a more aggressive casting checker that will detect what HP-UX does, and see if it throws anything up. Nicholas ClarkThread Previous | Thread Next