At 7:32 AM +0300 6/16/02, Jarkko Hietaniemi wrote: >Thanks, applied. At least in UNICOS this shouldn't be necessary, >though, the Friday snapshot worked 100%. May affect VOS.
The reason it caused a test failure with D_FLOAT is that the maximum value is 1.7e38, so squaring 1e20 gets you in trouble. If you were squaring 1e154 or so (which the test suite apparently does not currently do) you would see the same thing with other floating point formats that have a range similar to IEEE but without the denormalization capability. -- ________________________________________ Craig A. Berry mailto:[EMAIL PROTECTED] "... getting out of a sonnet is much more difficult than getting in." Brad Leithauser
