Re: vehicle to autoparts relationships

From: NENASHI, Tegiri <tnmail42_at_gmail.com>
Date: Thu, 23 Nov 2006 02:37:02 +0100 (CET)
Message-ID: <Xns9883D1BEBCC96asdgba_at_194.177.96.26>


"Aloha Kakuikanu" <aloha.kakuikanu_at_yahoo.com> wrote in news:1164244417.310045.85570_at_e3g2000cwe.googlegroups.com:

>
> NENASHI, Tegiri wrote:

>> "Aloha Kakuikanu" <aloha.kakuikanu_at_yahoo.com> wrote in
>> news:1164226982.397468.272020_at_b28g2000cwb.googlegroups.com:
>>
>>
>> > Contrary to what object people may say that assembling and
>> > disassembling things within a computer environment is not a big
>> > deal. It is certainly not a reason to dismiss relational approach.
>> >
>> > Now, there are some larger parts, ggregated from the smaller ones,
>> > for example, trigger assembly, or stock. How do we handle these?
>> > Simple, they are just sets:
>> >
>> > table AssemblyParts (
>> >     assemblyPartNo integer,    // informally it is a set#
>> >     partNo integer,                  // foreign key to Parts
>> > );
>> >
>> > Granted, some of the interested queries become set joins, so you
>> > have to be familiar with the concept of set join. Other than that I
>> > don't see any problems. Do you?
>>
>> You want to say the set containment join,  is not it ?  If you do, 
>> there are two problems: 1) the performance of set containment join;
>> 2) the set valued attribute, or relation valued attribute, is not
>> realised by all the databases. One can have a separate relation in
>> the place of the relation valued attribute, of course, but then how
>> does one reference the relation from AssemblyParts with
>> assemblyPartNo  ? 

>
> No nested relations.

Then how you create the schema ? Please show with tables: what is it that assemblyPartNo references ?  

> Can't one express set containment join in plain
> SQL
One can but it is slow.

>the same way one writes relational division query? Next, how many

> parts a typical vehicle has, pehaps 1000? Wouldn't relational division
> kind of query perform OK on such relatively small dataset (no matter
> how bad the execution plan is)?

It depends of the query: one can not really say without it.

>
Received on Thu Nov 23 2006 - 02:37:02 CET

Original text of this message