NoSQL, meh

Basically, he used to use Zope, and didn't like that you needed to instantiate an object to get it out of the DB. Then he complains his coworkers had some kind of ambiguous problem with this that only the author was smart enough to solve, then the author recognizes that this isn't at all a problem with modern NoSQL solutions (unless you want it to be).<p>Color me confused, what was the point of this diatribe? ZODB sucks? His coworkers suck because they only understand SQL? NoSQL might suck because ZODB sucks?<p>For the record, SQL is a beast, and far from intuitive. I hardly think most NoSQL is too hard for the masses (for that matter considering that NoSQL encompasses so many different techs its hard to say anything about NoSQL as a group), it sounds like no one at his shop wanted to really learn another DB since they could just ask this guy to help them out.

So, because he used a horrible "NoSQL" system years ago, the NoSQL concept now is going to be the same. Is it just me, or is the quality of HN links going down?

A good reminder for us who haven't been around as long, ideas and tech tend to come full circle.<p>Hopefully this time around, the tools are better and the limitations are well-known (and I think they are).<p>[edit] I think I misused the term "come full circle" -- I meant something more like oscillation.

I don't think he explains the issues very well even though he mentions some of the symptoms.<p>A common characteristic of NoSQL systems is that they don't use a normalised schema. That has certain scalability benefits and in a few cases it may be easier to understand. But it also means that there are going to be redundancies and that the database itself does not check consistency.<p>So not being able to use a generic tool to change some piece of data is not down to immature tools or lack of knowledge, it's a fundamental characteristic of denormalised data structures.