On 12/12/2009 9:54 AM, Stefan Ram wrote:

> jacob navia<(E-Mail Removed)> writes:

>> What I mean with this sentence is that when reading code about an

>> algorithm, it is easier to read if the error handling code is not

>> shown.

>

> I have another definition of »algorithm«: An algorithm is

> the implementation of some feature using the means of a

> specific target language. [...]
You're free to use words in your own way ("Who's to be

master?" -- H.D.), but your definition of "algorithm" is at

variance with the accepted meaning. That's likely to create

confusion and raise barriers to communication, just as if

you made an appointment to meet someone at noon, using your

own private definition of "noon" as "when the Moon rises."

But back to "algorithm:" Under your definition, there's

one "algorithm" for finding the GCD of two numbers in C,

another for C++, another for Basic, another for Java, ...

As a competent programmer, it seems you must learn a new

"algorithm" for every programming language, perhaps even for

every framework. That seems an awfully heavy cognitive burden.

In fact, under your definition "Euclid's Algorithm" is

an empty fiction, and ought to be something like "Euclid's

Recipe" or "Euclid's Way Of Going About It." There is

clearly an implementation-independent Something about this

GCD-finding approach; what do you call that i-i S, since

you use "algorithm" for something else?

Please answer before "noon."

--

Eric Sosman

http://www.velocityreviews.com/forums/(E-Mail Removed)lid