Re: Separate PK in Jxn Tbl?
Date: Tue, 29 Jan 2008 15:31:37 -0500
Message-ID: <u7OL0XrYIHA.4880_at_TK2MSFTNGP03.phx.gbl>
> You misunderstand my point. A great many people use surrogate keys as if
> they were surrogate addresses. They then use foreign keys that reference
> surrogate keys as if they were surrogate pointers.
Not sure to fully understand you on this sentence. However, if you mean by that that a surrogate key should never cross the boundaries of a database - with the inclusion of the interface as part of the database - then yes, theoritically you're right but practically - ie. in the real world with real budget - it's not necessary wrong to do it.
Some people consider databases as some kind of Gods and you shouldn't ask what the database can do for you but what you can do for the database. Personnally, I've never made a sacrifice - either monetary, animal, human or other - in the name of a database and all I'm asking of a database is what it can do for the client; never the other way.
However, like everything else, you must think about it and everytime you see a situation where a surrogate key is getting out of the database, there is a potential problem to look at.
-- Sylvain Lafontaine, ing. MVP - Technologies Virtual-PC E-mail: sylvain aei ca (fill the blanks, no spam please) "David Cressey" <cressey73_at_verizon.net> wrote in message news:8gHnj.525$4f.510_at_trndny08...Received on Tue Jan 29 2008 - 21:31:37 CET
>
> "Sylvain Lafontaine" <sylvain aei ca (fill the blanks, no spam please)>
> wrote in message news:%23Cj52$SYIHA.220_at_TK2MSFTNGP04.phx.gbl...
>> « But many of them seem to write as if contents as determined by
> ddress ».
>>
>> The content is not determined by the address and in fact, for those who
> are
>> using surrogate keys, the exact value of an address inside the database
> has
>> zero importance.
>
> You misunderstand my point. A great many people use surrogate keys as if
> they were surrogate addresses. They then use foreign keys that reference
> surrogate keys as if they were surrogate pointers.
>
> If this mimicry of pointers is pushed far enough, it can reduce a
> relational or SQL database to an inferior imitation of the old network
> databases. You lose the inherent advantages of content based addressing,
> but retain all the extra overhead.
>
>