Hacker News Reader

Comment by thom | Hacker News Reader

[flagged]

SQL itself doesn't generate any value, relational databases generate value. SQL is just a frontend for them. Anyway, your snark could be applied to _literally any change_. Are you angry that cars are replacing horses? "Horses generate such incomparable value that there's a steady supply of pro-car posts, like fumes from the vast ocean of their constantly boiling piss". Don't like the cotton gin? "Slave labor generates such incomparable value...". There's not really anything of substance in this kind of comment.

I think here applies very well this quote: All right, but apart from sanitation, medicine, education, wine, public order, irrigation, roads, the fresh-water system and public health, what have the Roman (SQL) ever done for us?

Not really. That quote is about who pioneered those innovations, not about who perfected them. No one disputes that SQL was first on the scene, we’re disputing whether it can be improved upon and whether the value derived from relational databases is a product of SQL specifically (hint: it’s not).

If, after 100 years of cars being available, everybody still rode horses and always got where they're going on time, and car people constantly blogged about it, this would be a great analogy.

I don't have a choice to use SQL. If I want to speak to a database, there is precisely one language available. Like Javascript on the web for so long.

This comment doesn’t make sense. There are 0 popular relational databases that support SQL and a dialect that addresses SQL’s shortcomings.

SQL as a query description language absolutely brings its own value.

We had a whole period of NoSQL because its difficult to shard SQL out across distributed databases, followed by people figuring out how to make SQL work on distributed databases, because SQL is really useful and people like it.

The whole period of NoSQL was particularly sad one. A couple great things emerged from it (Redis etc al) but for the most part it as all smoke and mirrors.

The worst was the amount of people wanting to be the next Ed Codd and being nowhere close to his mathematical background.

I'd call it a necessary phase rather than sad. People needed scalability fast, key-value stores are natural to shard, so they did that first and then we got better distributed DBs later. Even Prime Google was on bigtable for over a decade before spanner.

Reply on news.ycombinator.com