> affect the packages: Octave build fine but the test-suite displayed > problem inside the emulator, however the test-suite run on the same > binary on real hardware did not, so the package was correctly built."
Doesn't that still imply that compile-time constant folding could have been mis-executed, still leading to incorrect executables? (I'm recalling personal experience with a related situation with the IBM RT (back when it *wasn't* retro :-) and 3 floating point implementations, one of which was buggy - so there were runtime bugs on the broken hardware, but there were also mis-folded constants in anything [xdvi in particular] built on the broken hardware...) (I do favor using the fastest correct platform for builders, even if that's a virtual one - but it really does have to be correct...) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
