RT #77198: from_base truncation of large integers

[SineSwiper]

Still having any problems testing this branch or does everything look good?

[SineSwiper]

Hmmm...if I'm re-reviewing this code right, the possibility exists that this could break for big numbers, since it's trying to do math on a string. This should be:

return accurate_return( Math::BigFloat->new( $self->from_base($str) )->bneg() )
  if $nc && $str =~ s/^\Q$nc\E//; # Handle negative numbers

Where accurate_return() is that code piece on the bottom of from_base. I'll add a commit with that and a test for large negatives. Also, it looks like to_base negatives needs a quick big number check to make sure it can do -1*$num.

[SineSwiper]

My version of bbyrd has the latest commit.

[SineSwiper]

FYI, you might want to merge in this last change:

SineSwiper@6546f19

I figured out the nuances of Big*'s upgrade routine, and it should upgrade properly in these cases.

[kenahoo]

Thanks. I've merged it in locally (got a small conflict because of the last class-name patch), now I need to test, which means installing Dist::Zilla and all its dependencies, so 3.5 billion years later I should be good to go, assuming things test out okay.

[kenahoo]

I'm still getting a lot of failures, looks about the same as before. Here's the first failure:

t/01_basic_tests.t ..... 1/15
#   Failed test 'negative hex (-11) into decimal (-17)'
#   at t/01_basic_tests.t line 47.
#          got: '-17e+0'
#     expected: '-17'

I can commit this (to bbyrd branch) if you'd like?

[kenahoo]

Here's a little test that shows the root of the problem.

% perl -MMath::BigFloat -le 'print Math::BigFloat->new(11)->numify'
11e+0

[SineSwiper]

Bah, stupid BigMath bugs:

loubbyrd@noclab01:~$ perl -MMath::BigFloat -le "print Math::BigFloat->new(11)->numify"
11e+0
loubbyrd@noclab01:~$ perl -MMath::BigFloat -E 'say $Math::BigFloat::VERSION'
1.60

Z:\code\_others>perl -MMath::BigFloat -le "print Math::BigFloat->new(11)->numify"
11

Z:\code\_others>perl -MMath::BigFloat -E "say $Math::BigFloat::VERSION"
1.997

I was thinking of fixing it here, but you know what... 1.60 was made in 2002. Making a dep of a BigMath version that was updated sometime after the Nixon administration might be a better idea :)

Thoughts?

[kenahoo]

I think it's fine to depend on a later version where it's fixed, as long as that's available without upgrading perl. And that seems to be the case, so I'd say go for it.

[SineSwiper]

Interesting. The problem was only fixed in 1.997, so we'll need to make the dep be "Math::BigFloat >= 1.997". Version 1.993 still has the issue.

[SineSwiper]

Okay, I got back to this issue, rebased it, and fixed the problems in such a way that it would work on older versions of BigFloat. The last commit is SineSwiper/perl-math-basecalc@579656e676a, and that forked branch has the rebasing.