Re: Mixing OO and DB

From: Bob Badour <bbadour_at_pei.sympatico.ca>
Date: Sat, 22 Mar 2008 10:26:40 -0300
Message-ID: <47e50912$0$4034$9a566e8b_at_news.aliant.net>


frebe wrote:

> On 22 Mar, 00:23, Patrick May <p..._at_spe.com> wrote:
> 

>>frebe <freb..._at_gmail.com> writes:
>>
>>>On 19 Mar, 22:31, Patrick May <p..._at_spe.com> wrote:
>>>
>>>>frebe <freb..._at_gmail.com> writes:

[snip]

>> To give an example of decoupling, consider a clearing and
>>settlement system. The schema holds customers, accounts, positions,
>>books, trades, instruments, and other information related to the
>>clearing and settlement of financial transactions. Some of the
>>business logic is best expressed as operations on directed graphs
>>(eliminating cycles to create a directed acyclic graph, identifying
>>liquidity bottlenecks, selecting the transactions to settle, etc.).
>>These graphs need to be dynamically updated and are sufficiently large
>>that recreating them for every operation is prohibitively expensive in
>>terms of performance.

What an idiotic thing to suggest in the first place! Who cares if transactions have cycles? The only reason one would need to get rid of cycles is if one was stupid enough to treat the problem as a graph traversal in the first place.

> Do you claim that directed graphs could be represented as tuples in
> relations? Do you claim it doesn't exists a solution that only uses
> relations as data structure?

Frebe, if you are going to respond to these idiots, please give better answers.

[snip] Received on Sat Mar 22 2008 - 14:26:40 CET

Original text of this message