From: Jacob Fugal <lukfugl@...>
Date: 2009-03-07T13:16:17+09:00
Subject: [ruby-core:22710] Re: suggestions for float

On Fri, Mar 6, 2009 at 6:20 PM, Daniel Luz <dev@mernen.com> wrote:
> Again, what about something like Python's?
>
> �>>> str(0.1) � � � � �# equivalent to Ruby's #to_s
> �'0.1'
> �>>> repr(0.1) � � � � # equivalent to Ruby's #inspect
> �'0.10000000000000001'
> �>>> print 0.1 � � � � # print casts to string, just like Ruby's puts
> �0.1
> �>>> 0.1 � � � � � � � # the interactive interpreter uses repr, like irb
> �0.10000000000000001
> �>>> 2.1 - 3.0
> �-0.89999999999999991
>
> Keeping #to_s behavior unchanged would break very little existing
> code, and having #inspect return something that allows reconstructing
> the same object is generally a very nice thing, IMO.

I feel a lot more comfortable "fixing" Float#inspect to be more
verbose than I do Float#to_s. I'm against changing Float#to_s in the
ways proposed, but can accept changing Float#inspect.

That said, I do a lot of number crunching in irb; if I want some quick
calculator action, for instance, I just fire up irb and go at it. I,
personally, don't want to see a bunch of extra digits that I am
already aware of in the back of my mind but which aren't relevant for
me.

To accomodate both worlds, let's be sure to include a means to revert
to the old Float#inspect (without redefining it ourselves) that could
be included in .irbrc files. The default can be the verbose
Float#inspect, but easily overridden for users who like the old
behavior.

Jacob Fugal