http://cheerleader.yoz.com/2002/08/object-relational-persistence.html I came here to apologise and eat biscuits, and I'm all out of biscuits Fri, 25 Nov 2016 14:37:10 +0000 hourly 1 https://wordpress.org/?v=4.7.29 http://cheerleader.yoz.com/2002/08/object-relational-persistence.html#comment-26 Thu, 22 Aug 2002 23:01:19 +0000 http://cheertest.yoz.com/?p=42#comment-26 Are you talking about the JAWS type thing where the objects are turned into ‘real’ SQL rows? Sorry, I thought you meant serialisation.
Yes, there is more utility in that approach, but there are impedance matching problems… that’s what container managed entity beans in EJB try to do, but keeping schema and class in synch can be a problem, as can inheritance; I’ve not seen it done elegantly yet.
I’d rather just ditch SQL 🙂 SQL is the bane of my life!

]]> http://cheerleader.yoz.com/2002/08/object-relational-persistence.html#comment-25 Thu, 22 Aug 2002 14:37:52 +0000 http://cheertest.yoz.com/?p=42#comment-25 But this is about more than just using an RDBMS as a big storage vat for opaquely-serialised objects, which is useless. The point is to store this stuff in a database that’s good at querying, and in a way that translates to a proper schema rather than serialising. While these frameworks let you avoid SQL if you want to, they don’t mean that nobody can use it. Your heavy DB coders and admins can write (or use) big SQL report tools, while your appliation coders don’t have to touch that stuff.

]]> http://cheerleader.yoz.com/2002/08/object-relational-persistence.html#comment-24 Wed, 21 Aug 2002 17:48:19 +0000 http://cheertest.yoz.com/?p=42#comment-24 Our experience with this stuff is that it’s horribly horribly shit!
Basically, once you’re using an RDBMS as a place to store serialised objects, you’re better off with something like a raw DBM file. If you’re not really *using* that complex SQL data management layer between the DBM file and your application, then getting rid of it can produce large speed improvements and easier coding.

]]>