Front page | perl.perl5.porters |
Postings from May 2001
On Fri, Apr 27, 2001 at 10:55:53AM -0500, Jarkko Hietaniemi wrote: > On Fri, Apr 27, 2001 at 11:04:47AM +0200, H.M. Brand wrote: > > We're in stable water again. Thank you guys. > > HP-UX 11.00 stdio/perlio -DDEBUGGING -Duseperlio -Duse64bitall -Dusethreads -Duseithreads > > op/cmp................FAILED at test 0 > > Once Nick Clark gets his head wrapped around NaNs we may have > a solution for this. Is there a good way to generate 2147483648.0 4294967296.0 9223372036854775808.0 18446744073709551616.0 and any other 2**n and 2**(n-1) needed at configure time, without assuming that printf %g will correctly generate 20 digits, or that it can be done in integers? [note that I'd like MAX_INT + 1 so that it's a < test with large things and NaN falling on the "else" side] What I'd really like to do is generate 2.0 * 2.0 * 2.0 .... and let the compiler's constant folding sort it. However, just sticking the values for 32 and 64 bit IVs in a header and switching between them and an error message would probably work. > > lib/bigfltpm..........FAILED at test 351 Was there an answer for this one? Nicholas ClarkThread Previous | Thread Next