"Michael F. Stemper" <firstname.lastname@example.org> wrote in message news:email@example.com... > On 09/12/2013 01:54 PM, Julio Di Egidio wrote: >> "Peter Percival" <firstname.lastname@example.org> wrote in message >> news:email@example.com... >> >>> I've told you: the programmer should do what the specification says. >> >> You are wrong, better think writing a math library. > > What? When a programmer's writing a math library, s/he shouldn't do > what the specification says?
Sure, after s/he has written it... IOW, and for what is relevant to this thread (I do not teach software development in sci.math), not only PP's kind of suggestion just begs the question, it also reflects some limits in the scenarios that have been most discussed: my hint is to a math library as in that case it is quite obvious that one simply won't be able to "keep it under control". Of course, as others have already noted, one is expected to document that in his or her library 0^0:=1, but the fact remains that the burden of analysis (of consistency of the calculations performed) just escalates to the user, so even this indication only begs the question... Indeed, it is a straight mathematical question: what is 0^0 and why/why not with proof, and the rest just belongs to the usual fallacies.