Re: Types and "join compatibility"

From: paul c <toledobythesea_at_oohay.ac>
Date: Sat, 30 Jul 2005 23:34:08 GMT
Message-ID: <Q7UGe.69642$5V4.64590_at_pd7tw3no>


Marshall Spight wrote:

> ...
> My take: subtyping adds a huge amount of complexity to these
> issues and adds little of value. Subtyping is hugely valuable
> in OOP languages but I'm not convinced it adds much to relational
> languages. It's possible I'm wrong about the value added, but
> I'm certainly right about the complexity; it's worse than it
> looks.
> ...
> Again, my advice is, don't allow subtyping into your language.

although ttm seems to do that, it's not clear to me that 'relational languages' must forego whatever advantages subtyping might have if domain support is considered to be 'outside' the rm scope.

i'd like to see a restricted implemention of tutorial D that came with only 1 builtin type - that of table_dee (not even table_dum) but which would let you write your own domain support in whatever crappy language you could find or make a binding/linking arrangement for (i'd restrict it other ways too, such as avoiding the cardinality connotation of insert and delete). things would be chaotic i'm sure but they are that now anyway and it might make the more thoughtful pundits play the ball where it lay.

p Received on Sun Jul 31 2005 - 01:34:08 CEST

Original text of this message