Prevalence is cool

I think prevalence is cool. Why? Because it's such an "unthinkable" concept. It's almost tabu. I like the lightly smug attitude of the guys at Prevayler.org because they have actually created something based on this idea and are provocative enough for the RDBMS crowd to actually come up with good arguments.

Another reason I like this thought is because I think that RDMS's are one of the primary reasons for the relatively slow evolution, slow gain of oo development. When oo first arose it was predicted to be the next silver bullet. Nevertheless a lot of apps are made with the RDBMS as the dictative factor.

The constant need to persist data in relational data stores restricts the minds of programmers, and there are no doubt that an architectural sound, well designed app takes longer to develop than a VB'ish heavily db connected app. So the combination of programmers volountary or involountary use of RDBMS for persistance combined with time-to-marked requrements causes architecture and maintainability/changeability features to be down-prioritized.

So, find completely other ways to do stuff, the worst that can happen is that it actually sucks, but at least then we'll know;)

Update: bamboo.Prevalence is the .NET implementation of this project and stronglytyped has a cool view on the use of it.

Related to what Richard Caetano writes, it seems as a step on the way to separate snapshot data (in Prevalence) from pure warehouse data in a database. When you think about it this is actually how it works with databases today aswell; data has to be in memory in some shape or form to be accessed.

I think it's just a matter of time before solutions like Caetano described with completely transparent access to business objects will be reality.