On Dec 16, 12:08 pm, "frosty" <fros... (AT) bogus (DOT) tld> wrote:
Quote:
Sounds oddly familiar.
Tony opined:
It'll never work. Codd said so. |
He was actually a tad bit smarter than that, I think, at least until
he got caught up in the whirlwind marketing blitz known as the
relational database. Codd knew that you certainly COULD do that. It
is his disciples and then vendors who messed it up IMHO, not to
mention the SQL language.
Quote:
I was gonna post to cdt, but then I realized:
I don't care what those folks think. |
And you are, of course, a smarter man than I, Frosty ;-) but I am
finally to the point of agreement on this one now that somehow the
industry on the whole no longer seems to care. It might not be
exactly Pick, but the Pick approach to multivalues, use of two-valued
logic, use of a primary key (the relational model, in theory, has
candidate keys and not one selected as primary), variable length
fields (these days a lot of developers using rdbms's just make every
field 50 in length unless they have a reason to change it from that),
along with user-defined functions (virtual fields, computed columns,
calculated attributes all equal what we called I-descriptors in Prime
Information) are all back "in." Navigation replacing some of the set-
based logic has not found its way back in a big way, although perhaps
some tools help facilitate that too.
And don't forget, the "epoch date" (term used by Cache' for day 0 on
their calendar) for Pick gets a 40th birthday this New Year's Eve.
Databases have definitely had improvements during that time, but many
folks are returning to the simple and elegant model of data pre-dates
the relational model, whether they realize it or not.
Cheers! --dawn
Quote:
Appears
that Amazon doesn't, either.
--
frosty |