Dyalog Forums

I have been looking at column based databases for some time. The most obvious one for APLers is KDB+ from Arthur Whitney. Obviously there is also JDB and then there is the prototype database from Morten that you can obtain from the Dyalog site.
While generally looking around I found that Stonebraker, the Turin prize winner, father of Ingress, Postgres and then Vertica (a column database now owned by HP) has also now produced a new Array based database called SciDB. He even mentions APL in one of his papers about the reasons why he developed SciDB. Now sure his original work on SciDB was to capture scientific data but he appears now to extol Array databases as the way forward for IoT and even the financial markets that KDB+ addresses.
So I wondered whether APL missed a chance early on in its history -- was its ability to store arrays in files the way we should have gone instead of relational databases? I presume that Stonebraker has taken the idea further than just files of arrays but then he has the benefit of 30 years of the world's experience with relational databases?
I am always extremely impressed by what KDB+ can do, but has Stonebraker gone one step further?

When the fork occurred between J and K with K going in the direction of John Backus Turing lecture http://delivery.acm.org/10.1145/1290000 ... backus.pdf ( a prelude to Iverson's ) observation that matrices were just lists of same-length lists and really "recursive" lists was all that was needed , I went with the simplicity of K .

Thus the structure of CoSy , very much modeled on K , but built in open-to-the-x86 Forth is , if you will , "columnar" . CoSy's still more a GitHub level project at this time , but see the Twitter links on http://CoSy.com to the talk I gave at EuroForth , Vienna just before the Dyalog gathering where it was great to see so many old bright friends . Also linked is a comparison in APL , K , and CoSy of a quintessentially APL "age" problem Stephen Taylor mentioned when I got up to see him in London .