Re: Values have types ??

Leandro Guimarães Faria Corsetti Dutra wrote:
> On Sat, 06 Sep 2003 06:15:47 -0700, Costin Cozianu wrote:
>
>
>>Go back to your mathematical books and show me a single instance where 2
>>has an associated type specified for it. I bet you won't find it.
>
>
> Thing is, when you are talking Math you are usually assuming base 10.
>

That is, unless you are trolling on usenet :)

> But for the database, 2 is a symbol.

Skip that phase, assume it was parsed in 2 as in the number 2. All databases and their grandmothers take 2 as the number 2.

> For example, in your line of reasoning the database wouldn't know
> if that 2 is a number or a string.

Oh, please. The quasi-totality of programming languages adopts the following convention :

2	is a number
"2"	is a string
'2'	is the single character with the ascii code 50

> That is not so bad, as automatic type
> casting in this case would be pretty much straightforward. But what about
> 1011? You surely want to know if that's decimal, octal, hexadecimal or
> binary, don't you?
>

No, 1011 is in base 10 , obviously. For hexadecimal, binary, octal the language designer can design specific formats.

>
>
>>>>In the above case I'd propose that the MST is, well, {2}.
>>>
>>> That meaning? You see, the type is part of the meaning...
>>>
>>
>>Meaning the set with only one element, 2. The standard notation for it is
>>{2}.
>
>
> Thank you. So what?
>
>

So as I said, for the whole theory to preserve some kind of consistency, we end up with

	all values x
	have the "MST" {x}

And that will make it a pretty much useless concept. Specialization by constraint is a hopelessly naive approach.

Not to mention that it runs contrary to more than 4 decades of theoretical research and praxis in the specialized field of mathematics and CS that is called type theory -- that is a whole domain of mathematics unknown to D&D, and consequently to their ignorami followers.

Therefore all these pet theory of "type inheritance" (horrible dictu) are just pet theory, that are inadequate for either theoretical or practical purposes, yet both you, Bob and a few argue as if you were talking about the eternal mathematical truth. Just arguments from ignorance.

Useful theories usually talk in terms of algebras and their elements ( types are algebras ) and functions are operations on those algebras, or more abstractly types are objects categories and computations are arrows. More recently Goguen proposed to consider the stronger types are theories. Types = sets has long been deemed a useless approach. Received on Sun Sep 07 2003 - 02:23:59 CEST