Re: [boost] metactrl header [compile time asserts]

From: David Abrahams (abrahams_at_[hidden])
Date: 2000-07-28 12:19:44

----- Original Message -----
From: "Borgerding, Mark A." <MarkAB_at_[hidden]>

> I am not suggesting that a programmer would be able to apply the same
> scrutiny and rigor to a piece of code as a mathematician does to a proof.
> If the code is laid out thoughtfully, then the successful compilation of
the
> program could be considered analogous to a mathematical proof, with most
> assumptions checked at compile time.  However, there are generally very
> basic concepts that we assume when programming: the computer will not lose
> power, memory will not have bit errors, filesystems will not be corrupted.
> We make these assumptions, not because they are obviously true, but
because
> without assuming *something*, we cannot make functional programs.  Even if
I
> were to write a program that checked a filesystem's integrity, I would
> probably need to assume the memory in the computer was reliable.

Right. These are postulates. When you say "I postulate X" you are saying
up-front, "I am assuming X is true; I am not going to make any attempt to
check it; I think it's simple enough that you should be willing to accept it
without checking it". That's why I think it's an inappropriate name for a
construct which does checking.

-Dave