On Apr 23, 2016, at 19:43 , David Bennett <davidb@pfxcorp.com> wrote:
> owner@postgresql.org] On Behalf Of Eric Schwarzenbach
>
>>>> If I had a few $million to spend in a philanthropical manner, I would
>>>> hire some of the best PG devs to develop a proper relational database
>> server.
>>>> Probably a query language that expressed the relational algebra in a
>>>> scheme-like syntax, and the storage model would be properly
>>>> relational (eg no duplicate rows).
>
> If there were someone to pay the bills, would you work on it?
Yes, to the extent that my skills were useful (I'm good at logic, but I've always gravitated to higher-level languages,
soif it was C, I'd write documentation or something).
>>>> It's an enormous tragedy that all the development effort that has
>>>> gone into NoSQL database has pretty much all gotten it wrong: by all
>>>> means throw out SQL, but not the relational model with it. They're
>>>> all just rehashing the debate over hierarchical storage from the 70s.
>>>> Comp Sci courses should feature a history class.
>>>>
>>>> It's a bit odd to me that someone isn't working on such a thing.
>
> Several people are, but without the few $million…
I hear you. I started this discussion because much as I utterly adore what the miracle workers behind Postgres have
managedto create from the sow's ear that is SQL, I'm always aware of just how much better the world could be. And data
storageis so central to everything that happens. Now is the time, with this NoSQL movement afoot, to provide something
thatbreaks from SQL in the *right* way.
I would love to be able to choose an eventually-consistent store that syncs with minimal effort with a local store in a
handheldapp, or any of these great storage options that the NoSQL folks are doing, but giving up relations is *such* a
highprice to pay, particularly when I know that it just isn't necessary.
If someone was working on such a thing in a language that is less tedious than C (Go or Scheme or Haskell or Clojure,
anyone?),I would happily contribute.
My ideal data store is something like a top-level relational engine with pluggable stores, pluggable synchronization
etcoptions, pluggable languages. Pretty much the philosophy that has driven Postgres from the beginning. If such a
thingcould be produced and became popular, it would almost add measurably to GDP, I think, such could be its impact if
doneright.
Anyway, I hope everyone has found this as interesting a discussion as I have. I'll dig more deeply into Andl and watch
itsprogress. And if anyone here knows of a non-SQL relational database project worth noting, please do so here.