Re: Is relational theory irrelevant?
Date: Wed, 19 Nov 2003 19:16:09 -0000
Message-ID: <bpgfo3$1vc2$1_at_gazette.almaden.ibm.com>
"Bob Badour" <bbadour_at_golden.net> wrote in message
news:HfudnVql5P9mNCeiRVn-sA_at_golden.net...
> "Paul Vernon" <paul.vernon_at_ukk.ibmm.comm> wrote in message
> news:bpe6mq$1n9m$1_at_gazette.almaden.ibm.com...
> > "Bob Badour" <bbadour_at_golden.net> wrote in message
> > news:ToednURmpdK7Dyei4p2dnA_at_golden.net...
> > > > Interesting. I didn't intend to allege that side-effects are bad.
They
> > > > are a reminder that SQL has to deal with the real world.
> > >
> > > Side-effects are bad. I suggest your example was also bad. Sending
> emails
> > is
> > > not data management. Recording the emails sent is data management. A
> > > database management system has a role and a function. Sending emails
is
> > not
> > > its function.
> >
> > No, but if you want to get rid of logical transactions, then sending
> emails
> > is and example of an activity that needs to hook into the physical layer
> of
> > the database, which would have the consequence of users of the logical
> > database experiencing that the DBMS does indeed send emails. :-)
>
> Your assertion seems non-sequitur to me. Can you explain why you think
this
> requirement exists and is caused as asserted above?
Sending emails is, I suggest, and example of an integration with an external system, and may well therefore require a 'hook into the physical layer'. BTW by hook, I really mean some (limited) transaction model in some layer beneath the logical model.
For example, the sendmail agent that writes to the Email_Send_Result table (in my prev post) should use a transaction to tie the act of sending an email with the database update to reflect the result. A system crash between the email being sent and the database being updated, should not result in incorrect information about the send result being stored in the database.
Such transactions should however, neither be visible nor be available to users of the logical model.
Regards
Paul Vernon
Business Intelligence, IBM Global Services
Received on Wed Nov 19 2003 - 20:16:09 CET
