From: Stefan Schmiedl <s@...>
Date: 2001-12-11T04:25:00+09:00
Subject: [ruby-talk:28129] Re: John Roth dolt (  Re: A challenge to proponents of Unit Testing. )

Joe Pallas (pallas at cs dot stanford dot edu) (2001-12-11 03:40):

> Since you've implied that your function will always throw an exception
> for bad arguments, let me point out that the MTFB of your function is
> zero if given either of the inputs (5, 5, 11), which it will incorrectly
> report to be an isosceles triangle, or (1, 3, 7), which it will
> incorrectly report to be a scalene triangle.
> 

I would definitely prefer to factor deciding the validity into
a separate method which is tested separately.
Classification is so much easier when you know the species.

Besides, I understood the message which started this discussion,
to classifiy a triangle described by its sides. So it is up to
the environment to decide whether (5,5,11) is passed into
classification (would be possible in non-Euclidean geometries I think).
Or if Euclidean borderline cases as (5,5,10) are considered as triangles
or not.

Seeing it this way makes for smaller methods which I can easier
understand. Of course, your mileage may vary.

s.
-- 
Stefan Schmiedl
EDV-Beratung, Programmierung, Schulung
Loreleystr. 5, 94315 Straubing, Germany
Tel. (0 94 21) 74 01 06
Public Key: http://xss.de/stefan.public

shhhh ... I can't hear my code!