Re: K.I.S.S. Factor (Re: Numerics)

On 26/03/2005, at 11:11 AM, Cyril ADRIAN wrote:
> As soon as conversions are implemented, people are expected to want to 
> convert anything into everything. There is the beginning of real 
> trouble.

The experience of conversions in ISE Eiffel (which has had them for a 
couple of years) suggests otherwise.

If anything, the ISE Eiffel team has been a bit too cautious in 
providing conversions between Eiffel and .NET classes. I've been caught 
by surprise a couple of times by the lack of a conversion that would 
have provided an intuitive mapping. I've yet to encounter a conversion 
that caused me any confusion whatsoever.

The ISE Eiffel experience is confirmed by the C# experience, which has 
also had conversions for several years amongst a much wider community 
of programmers. Writing conversions in C# is considered a fairly 
advanced technique, so few C# programmers attempt it. When they do, 
they tend to do so responsibly, resulting in behaviour that is 
unsurprising to other programmers who use those classes.

Programmers use conversions rarely, responsibly and cautiously, in my 
experience.

- Peter Gummer