Re: Objects and Relations

From: Bob Badour <bbadour_at_pei.sympatico.ca>
Date: Fri, 16 Feb 2007 23:01:01 GMT
Message-ID: <NIqBh.7415$R71.110825@ursa-nb00s0.nbnet.nb.ca>

Kevin Kirkpatrick wrote:

> On Feb 14, 6:46 pm, "David BL" <davi..._at_iinet.net.au> wrote:
>

>>On Feb 15, 12:36 am, "Kevin Kirkpatrick" <kvnkrkpt..._at_gmail.com>
>>wrote:
>>
>>>On Feb 14, 7:12 am, "Cimode" <cim..._at_hotmail.com> wrote:
>>
>>>>[Snipped crap from Keith Duggar]
>>
>>>>>Nevertheless, for the benefit of other readers if nothing else, let me
>>>>>state that Keith has hit the nail on the head.
>>
>>>>No.  Keith has just once shown more misunderstanding of RM and
>>>>arrogance.
>>
>>>>>1) Lego blocks are /not/ the same. They always have a different
>>>>>location attribute.
>>
>>>>BS.  LegoBlock are indeed the same if they are to be a relation.
>>
>>>>Stating that location attribute is a primary natural key for a
>>>>LegoBlock is arbitrary and does not reflect the reality of
>>>>LegoBlocks.  Imagine a M:N cardinality between LegoBlock and Location
>>>>Entity.  In that case LocationEntity has a natural identifier (X, Y,
>>>>Z).  The natural identifier of a lego block could be a combination of
>>>>its shape, color but certainly not its location.
>>
>>>>>2) Hence their x,y,z position attribute always identifies them.
>>
>>>>False.
>>
>>>>>3) However this identifer is very hard to record and keep track of,
>>>>>even though it exists.
>>
>>>>In the mind of people which pointer's obsessed maybe...
>>>>There is nothing complex about the identifier of a location is a
>>>>concatenation of XYZ and a referential.
>>
>>>>>4) So we represent it with a surrogate identifer (which is hence just
>>>>>an 'untrackable' attribute or, as others refer to it, an unfamiliar
>>>>>attribute).
>>
>>>>In other words, Surrogate identifier has only the advantage of being
>>>>*familiar*.
>>
>>>>>5) OID's are hence not needed, and everything is just a value, as it
>>>>>should be.
>>
>>>>OID = Implementation of Surrogate Key IF AND ONLY IF
>>>>--> unicity contraint has been implemented on the implementation of
>>>>the natural key
>>>>--> unicity constraint implemented on the OID
>>>>--> contraint of cardinality implemented 1:1 between the
>>>>implementation of the natural key and the OID....
>>
>>>>In other words, almost never happens...
>>
>>>>>6) When I finally understood this (and it was a hard mental slog to
>>>>>shake off my OO mindset) I found it quite revelationary to my
>>>>>perspective on data management.
>>
>>>>Guess we all went through this at some point in time...
>>
>>>>[rest snipped]
>>
>>>I have to agree with Cimode - Keith's contribution seemed a little...
>>>off.  Assuming lego blocks are of the simple cuboid variety with
>>>standard male and female connectors, a configuration can be expressed
>>>as tuples in the following relation (asterisks indicate the natural
>>>key columns):
>>
>>>CONNECTION  (male_color*, male_thickness*, male_num_cols*,
>>>male_num_rows*, male_column*, male_row*, female_color*,
>>>female_thickness*, female_num_cols*, female_num_rows*, female_column*,
>>>female_row* , rotational_angle)
>>
>>How can this work?  There is nothing to distinguish the individual
>>Lego blocks.  Your key for a connection only makes use of the /type/
>>of Lego block (parameterised by colour and shape).

>
> Damnit - for the third time in as many days, I will try to send a
> response. This time I'll type in notepad first, ensuring that it will
> work.
>
> Anyway, I see I've treated the problem quite poorly; in fact one
> cannot express a configuration without reference to individual Lego
> blocks. I apologize for the confusion, and moreso to Kieth for
> dismissing what was, in retrospect, a valid point.
>
> However, there is still no need to model individual Lego blocks as
> real-world distinct entities (e.g. with a global Block_Id uniquely
> assigned to pieces across all configurations); one merely needs to
> identify Lego blocks according to their role within a configuration.
> This can be done with a compound surrogate key, e.g. arbitrarily
> numbering pieces within each configuration; or with a meaningful
> natural key, such as the x, y, z distance (in integral "standard Lego
> units") of the leftmost, foremost, bottommost corner of each piece
> from intersection of the leftmost, foremost, and bottommost edge of
> the overall configuration.
>
> I believe a full configuration could than be specified as:
>
> Config_pieces (config_name*, x*, y*, z*, color, width, depth, height,
> orientation).

If you have x, y, z, deltaX, deltaY, and deltaZ you won't need orientation. The location of a piece is actually an interval. Received on Fri Feb 16 2007 - 17:01:01 CST