On Friday, 2 August 2013 12:28:51 UTC+8, CGAT wrote:
> My understanding of == is that it is intended to establish numerical
> equivalence
> across types. But I think that basic contract fails with BigDecimal. For
> instance,
>
> (== 1M 1.0M) ; => false
>
> because the scale properties of these numbers are different. So then of
> course:
>
> [(== 1 1N 1.0) (== 1 1N 1.0 1M) (== 1 1N 1.0 1.0M) (== 1 1.0 1N 1.0M)]
> ; => [true true true false]
> and
> [(== 1.0M 1.0) (== 1.0 1) (== 1 1N) (== 1N 1.0M)]
> ; => [true true true false]
>
> I find your lack of transitivity (and commutativity) ... disturbing.
>
> The issue is that there are two notions of equality for BigDecimal,
> equals and compareTo, where equals compares value *and* scale while
> compareTo compares numerically.
>
> The other numeric types use equals for equivalence, quite reasonably.
> But in class BigDecimalOps in clojure/lang/Numbers.java, I propose
> that
>
> public boolean equiv(Number x, Number y){
> return toBigDecimal(x).equals(toBigDecimal(y));
> }
>
> should be
>
> public boolean equiv(Number x, Number y){
> return toBigDecimal(x).compareTo(toBigDecimal(y)) == 0;
> }
Ouch, is that really what Numbers.equiv is doing?
No wonder some of our idiomatic numerical code is so slow - a coercion to
BigDecimal is really expensive!
It really needs instanceof checks and branches to handle the different
cases efficiently. I'll have a go at a ticket/patch that addresses this.
--
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to [email protected]
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
