Float precision and FIT


#1

Hello,

I am trying to get the Division example on the FIT wiki
http://fit.c2.com/wiki.cgi?WhatAndHowTalk

to run under Ruby and the Fit framework. I have 2 questions. First I
get red results on last 3 tests because the precision in the HTML
table is only up to 8 digits, but Ruby is calculating 15 digits.

And second, even after I adjust the HTML to match the results, they
till don’t equal. I am thinking this is because of ‘==’ vs. eql? on
floats.

Is there a way coerce a Ruby float to an aribitray precision via
truncate or round?

Thanks


#2

Do you have a FIT fixture for running Ruby ?

j

On 11/28/05, Ed Howland removed_email_address@domain.invalid wrote:

And second, even after I adjust the HTML to match the results, they
till don’t equal. I am thinking this is because of ‘==’ vs. eql? on
floats.

Is there a way coerce a Ruby float to an aribitray precision via
truncate or round?

Thanks


“Remember. Understand. Believe. Yield! -> http://ruby-lang.org

Jeff W.


#3

On 11/28/05, Jeff W. removed_email_address@domain.invalid wrote:

Do you have a FIT fixture for running Ruby ?

j

Well, I have the Fit framwork d/l from the aforementioned Wiki. There
is apparently another
project in CVS on Rubyforge. Haven’t checked it out yet.

Jeff W.

Ed


#4

Ah.

If I remember correctly, there once was rumor that Dave T. was
working
on a FIT-like framework for Ruby …

I wonder whatever happened to that project…

I’ve written a little ruby replacement of FIT … I’ll probably post it
if
we don’t find that someone else has already done the work.

Haven’t taken the next step and built it into a wiki like the FITnesse
project …

j.

On 11/28/05, Ed Howland removed_email_address@domain.invalid wrote:

Jeff W.

Ed


“Remember. Understand. Believe. Yield! -> http://ruby-lang.org

Jeff W.


#5

On 11/28/05, Jeff W. removed_email_address@domain.invalid wrote:

Ah.

If I remember correctly, there once was rumor that Dave T. was working
on a FIT-like framework for Ruby …

I wonder whatever happened to that project…

I think that is the one I d/l from the wiki. Not sure though.

I’ve written a little ruby replacement of FIT … I’ll probably post it if
we don’t find that someone else has already done the work.

Haven’t taken the next step and built it into a wiki like the FITnesse
project …

There is a version of RubyFIT in CVS on RubyForge. I tried it out and
it ran the Simple example (eg.Division) right out of the box. Solves
the precision problem by using a type adapter. Looking forward to how
that code addresses the 5 digit precision issue in Java v. Ruby.

So take a look at that one. If yours is better/different somehow, go
ahead and post it.

http://rubyforge.org/projects/fit

j.

Ed


#6

On Nov 28, 2005, at 10:07 AM, Ed Howland wrote:

And second, even after I adjust the HTML to match the results, they
till don’t equal. I am thinking this is because of ‘==’ vs. eql? on
floats.

No. Float precision.

Is there a way coerce a Ruby float to an aribitray precision via
truncate or round?

For testing float equality do what Test::Unit’s assert_in_delta does.


Eric H. - removed_email_address@domain.invalid - http://segment7.net
This implementation is HODEL-HASH-9600 compliant

http://trackmap.robotcoop.com