develooper Front page | perl.perl5.porters | Postings from July 2015

Views on changing Math-BigInt OO design

Thread Next
From:
Peter John Acklam
Date:
July 28, 2015 08:39
Subject:
Views on changing Math-BigInt OO design
Message ID:
CAEvoY4178u58D5Tmw57aFhP31G3DeknLucsBmJjLbnNcfBuGAQ@mail.gmail.com
I want to change the OO design of the Math-BigInt distribution, but before
I do so, I would like your opinion.

Current implementation:

Currently, Math::BigInt is the superclass of Math::BigFloat, which in turn
is the superclass of Math::BigRat. This OO design doesn't make sense. None
of Math::BigInt, Math::BigFloat, and Math::BigRat have an internal
structure which is a superset/subset of any of the others. Methods like
band(), bor() etc. make sense on Math::BigInt objects, but don't make sense
on Math::BigFloat objects and Math::BigRat objects, who inherits the
methods band(), bor() etc.

TELS, who introduced this design, seemed to be aware of the problems, and
as a sort of workaround let Math::BigFloat->isa("Math::BigInt") return
false. Ditto for Math::BigRat. This doesn't solve the real problems and
causes problems (e.g., endless recursion) when creating subclasses.

New implementation:

I want to introduce Math::BigNum as a virtual parent class, and let
Math::BigInt, Math::BigFloat, and Math::BigRat be subclasses of
Math::BigNum. Math::BigNum will implement the very few methods that
Math::BigInt, Math::BigFloat, and Math::BigRat have in common as well as
some internal utility functions.

Would this be OK?

Regards,
Peter John Acklam
(current maintainer of Math-BigInt, Math-BigRat, and bignum)

Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About