Front page | perl.perl6.internals.bignum |
Postings from November 2001
Re: BIGNUM memory and questions
November 27, 2001 14:40
Re: BIGNUM memory and questions
Message ID: firstname.lastname@example.org
-----BEGIN PGP SIGNED MESSAGE-----
On 27-Nov-01 Uri Guttman tried to scribble about:
> T> Next question: Why BCD? Do you think this is better, or do you
> T> know? (E.g. why was that decision made. I am asking because there
> T> are differences and I want to make sure that the decision is done
> T> on research, not on "Uh, we thought it would be better" ;)
> bcd is more accurate which is a prime reason.
This is untrue. Neither binary nor BCD are more or less accurare. Otherwise
Math::BigInt lib => 'Pari' would not work 100% as Math::BigInt;
I guess we are talking about different things here!?
> it will allow for easier conversions to/from strings.
This is true. But it will make for very slow multiplications and things
like 2 ** something. (I bet you know that ;)
> It shouldn't be much slower as bcd algorithms are done with word wide
It will be _MUCH_ slower for some cases. Heh, want some benchmarks? ;)
> it is less dense but that is not a major disadvantage.
You are probably talking about memory consumption here? I agree, memory
shouldn't be the issue.
> T> (Yes, I have thought about Math::BigInt::Cached, which would allow
> T> a dual-usage if BCD and binary math at the same time. Currently I
> T> have no time to finish it, that comes after v5.8.0)
> >> this will allow for precise decimal math which is desired.
> T> This would be able with binary packages as well, as you can see by
> T> use Math::BigInt lib => 'Pari';
> T> use Math::BigFloat;
> those packages are so bad it is hard to believe.
Thank you for bashing my work.
> i actually am looking at bigfloat internals now to find an odd bug for a
> poster on usenet.
At which version are you looking? And what is the bug so that I can fix it?
> i wouldn't use that code for anything.
I really hope you are writing about the v0.01...
> T> Math::BigInt and Math::BigFloat will work 100% as they do with a
> T> decimal math package under the hood (which is the default).
> they are in perl4.
They are in Perl5 - not perl4. Please see http://bloodgate.com/perl or
Please look up the newest version, it has come a long way...
> and poorly written and slow.
I really hope you are not writing about my work. Otherwise I stop talking
right now and head back to the coding-board.
> not yet. that is part of what needs to be written. you obviously have
> not even looked at what i have so far.
See at bottom.
> and i have done a bcd math package before also. this is not new
> territory for me.
> T> Hope we can talk soon, because I see lot's of wasted time and I
> T> want to prevent this ;)
> then help out with code.
> T> Anyway, don't take that personal!
> you seem to. instead of being such a naysayer and pushing perl based
> solutions etc. why don't you work on the code base i have started?
Because I have my hands full of Math::BigInt - sorry, but I don't have time
for another project. Apart from that my C is rusty, I know next to nothing
of Parrot (last thing I remember was that it was an April's fool joke) and
besides, I like to know beforehand why and what I code. Me heading into your
project would be chaos.
And I don't push "perl-based" solutions. I am well aware that Perl is slow.
But it already has all the logic, tests, documentation and features that
one possible can want, so you could borrow something from it.
> everyone. and get the current tarball at stemsystems.com/bignum.tar.gz
Will check it out later.
perl -MDev::Bollocks -e'print Dev::Bollocks->rand(),"\n"'
preemptively develop front-end communities
http://bloodgate.com/perl My current Perl projects
PGP key available on http://bloodgate.com/tels.asc or via email
-----BEGIN PGP SIGNATURE-----
-----END PGP SIGNATURE-----