Re: Bug#130476: libmath-numbercruncher-perl: fails due to bigfloat?
Hi Neil,
On Tue, Jan 22, 2002 at 03:35:40PM -0800, Neil T. Spring wrote:
> Package: libmath-numbercruncher-perl
> Version: 4.05-1
> Severity: important
>
> evil:~/scripts/viz> ./rout95.pl
> Can't locate object method "copy" via package "Math::BigFloat" (perhaps you forgot to load "Math::BigFloat"?) at /usr/share/perl5/Math/NumberCruncher.pm line 982.
> Compilation failed in require at ./rout95.pl line 2.
> BEGIN failed--compilation aborted at ./rout95.pl line 2.
> evil:~/scripts/viz> head -3 ./rout95.pl
> #!/usr/bin/perl
> use Math::NumberCruncher;
>
> line 982 of NumberCruncher.pm is:
> my $max_econst = $_e_->copy();
>
> this should be pretty easy to recreate if it is indeed
> a problem. this script compiles and runs fine with 4.02-1
> on an otherwise similar machine.
Oh dear. The package now relies on Math::BigFloat (not part of Perl, at
least not 5.6.1) as well as version 1.49 of Math::BigInt -- which is newer
than the one shipped with 5.6.1.
What held me back was the following comment in the README for Math::BigInt:
This package contains new versions of Math::BigInt and Math::BigFloat,
so please do NOT make 'make install', except when you do no longer want your
old modules or know what you are doing.
Do you think I should enforce these?
I saw Math::NumberCruncher as a convenience package providing some useful
Math routines. In its current state it requires to overload a lot of the
tried and test Perl Math guts. I am not sure if I want this.
However, as you discovered, some operations do require it. I had noticed in
the last few release that 'make test' failed on some ops for this very
reason, but was still hesitating to enforce a new Math::BigInt.
Comments from your end? Would you be willing/able to downgrade to a 4.02
version? Or would you rather see Math::BigInt override your Perl defaults?
I am CCing the Debian Perl list to see if someone else has insights.
Thanks, Dirk
--
Good judgment comes from experience; experience comes from bad judgment.
-- F. Brooks
Reply to: