scsh-users
[Top] [All Lists]

Re: Is s48_extract_double actually useful?

To: Felix Klock <felix_klock_ii@mac.com>
Subject: Re: Is s48_extract_double actually useful?
From: Martin Gasbichler <gasbichl@informatik.uni-tuebingen.de>
Date: Tue, 17 Dec 2002 13:23:18 +0100
Cc: scsh-news@zurich.ai.mit.edu
>>>>> "Felix" == Felix Klock <felix_klock_ii@mac.com> writes:

Felix> Interesting.  That seems to have fixed the problem.  I had already
Felix> come up with a workaround, but this will be a bit nicer to use.

Felix> So what is the deal; is the default inexact arithmetic "better" than
Felix> using doubles internally?  Somehow I doubt it, unless it is using
Felix> bignums so it can claim more accuracy than doubles.  But once you're
Felix> using inexact numbers, I think that you've made it clear that you
Felix> don't care so much about accuracy...

Felix>   but hey, what do I know... I'll try to find the docs on the
Felix> floatnums module.

I don't think there is any documentation available on this, apart from
a small comment in scheme/more-packages.scm:

;; floatnums            ;Still don't print correctly

I think this is also the reason why Scheme 48 doesn't use them by
default. The implementation you get by default is provided by the
package innums, which are simply exact numbers which look (print) like
inexact numbers. This is of course much slower.

BTW: scsh uses the floatnums package by default.

-- 
Martin

<Prev in Thread] Current Thread [Next in Thread>