Re: OO and relation "impedance mismatch"

From: Bernard Peek <bap_at_shrdlu.com>
Date: Sat, 16 Oct 2004 18:59:19 GMT
Message-ID: <s$KWpET71WcBFwY5_at_shrdlu.com>


In message <5qadnW5UX_5ni-3cRVn-pA_at_comcast.com>, Laconic2 <laconic2_at_comcast.net> writes
>
>"Bernard Peek" <bap_at_shrdlu.com> wrote in message
>news:RV4eU7bFGAcBFwcf_at_shrdlu.com...
>> In message <iz%7d.409134$8_6.338269_at_attbi_s04>, Marshall Spight
>> <mspight_at_dnai.com> writes
>>
>>
>> >Limiting classes to only be applicable to relational domains would be
>> >hamstringing them.
>>
>> I've been thinking about this impedance mismatch for a while. One of the
>> things that I think contributes is the assumption that classes are
>> prescriptive rather than descriptive. That's a given if you approach the
>> problem using the existing OO languages that I'm aware of. The impedance
>> mismatch seems to be between OO languages and OO data storage systems.
>>
>> I'm visualising a system where a piece of code might be something like:
>>
>> CREATE CLASS REDTHINGS AS SELECT COLOUR = "RED"
>>
>> Thus classes would all be created dynamically as sets of objects meeting
>> a selection criterion.
>
>One of the things that I'm perplexed about is why the people who think about
>this impedance mismatch focus so much on the "class".
>
>The few formal descriptions of object oriented modeling I've seen say that
>the "class" is an OPTIONAL construct in object oriented modeling, while
>the "object" is fundamental. Maybe I'm missing something. Can you clarify?

That's why I think the impedance mismatch is within the OO community. The languages that OO programmers use seem to insist on classes, so people keep trying to implement them in OO databases. I'd be interested to see what would happen if a COBOL programmer designed an OO database.

-- 
Bernard Peek
London, UK. DBA, Manager, Trainer & Author. Will work for money.
Received on Sat Oct 16 2004 - 20:59:19 CEST

Original text of this message