Re^3: Small Perl 6 discoveries III, Ints

Replies are listed 'Best First'.
Re^4: Small Perl 6 discoveries III, Ints by syphilis (Archbishop) on Sep 29, 2017 at 05:16 UTC
`> 1.000000000000001e0 - 1e0 0` [download] Certainly can't reproduce that one on perl5 ;-) It looks buggy to me. It appears that 1.000000000000001e0 has been rounded to 1.00000000000000e0 prior to doing the subtraction. Though it's also possible that 0.000000000000001e0 has been rounded to 0 when the value is printed out. Does perl6 actually assert that `1.000000000000001e0 == 1e0` ?? Of course, "double" builds of perl5 will round 1.000000000000001e0 to 1e0 prior to printing it out: `C:\>perl -le "print 1.000000000000001e0;" 1` [download] but that's different to what appears to be happening in the given example. (And even though perl5 has printed out "1", it knows damn well that `1.000000000000001e0 != 1e0`) Cheers, Rob	[reply] [d/l] [select]
Re^5: Small Perl 6 discoveries III, Ints by holli (Abbot) on Sep 29, 2017 at 05:56 UTC
This too is due to the exponential notation always generating `Num`s instead of `Rat`s. So if you substract a value with rounding errors minus the same value without rounding errors, what you have left is the rounding error. holli You can lead your users to water, but alas, you cannot drown them.	[reply] [d/l] [select]
Re^6: Small Perl 6 discoveries III, Ints by Anonymous Monk on Sep 29, 2017 at 06:05 UTC
The rounding errors aren't the problem. It's the inconsistency of the rounding errors that's worrying us.	[reply]
Re^5: Small Perl 6 discoveries III, Ints by Anonymous Monk on Sep 29, 2017 at 05:46 UTC
`> my $a = 1.000000000000001e0; my $b = 1e0; $a - $b 0 > $a - 1 1.11022302462516e-15 > $b - 1 1.11022302462516e-15 > $a.WHERE 4405787632 > $b.WHERE 4405787632 > 1.000000000000001e0 == 1e0 True` [download]	[reply] [d/l]
Re^6: Small Perl 6 discoveries III, Ints by syphilis (Archbishop) on Sep 29, 2017 at 07:40 UTC
`> 1.000000000000001e0 == 1e0 True` [download] Oh dear, that is awful. Someone should inform the perl6 developers of that behaviour - assuming, of course, that they haven't yet fixed it. Cheers, Rob	[reply] [d/l]