What EC2 instance type would you recommend for running a Datomic Peer ? (use case: a 4G-RAM, OLTP web server)

I've been running mine on t2.medium so far but I don't know how to balance compute, memory and I/O specifically for a Peer

note that I do mean a Peer, not a Peer Server

we’re on c4.xlarge

i have a query like :

[:find   (pull ?tx [:db/txInstant]) (pull ?e [* {:order/items [* {:item/part [*]}]}])
          :in $
          :where [?e :order/number _ ?tx true]]

works fine but returns array of tuples [instant, order-object]

how can i rewrite the query so the txinstant is ‘merged’ into the entity?

you can pull in a where clause

but not sure if that's worth the trouble (instead I just map after querying)

how would a pull in where look like? sounds new to me...

map definitely sounds easier though 🙂

@ezmiller77 I’ve added a new page to the docs that covers getting a local dev setup running: http://docs.datomic.com/dev-setup.html

Given the recent discussion about schema lifecycle in here, I wanted to mention the blog post from Stu Halloway from today: http://blog.datomic.com/2017/01/the-ten-rules-of-schema-growth.html

I'd love to know, under Stu's recommendations, how I can break up a field? I have a field which is currently stored as edn strings, and are converted as the entity is read. Now those attributes need querying, so I want to deprecate that field. But how can I ever (from a query position) trust that field is entirely gone? Currently we do things like get all X, read their edn and filter (goodbye, optimizations from datomic!) Must I always keep this code around which is extremely unperformant in order to never break backwards compatibility?

Interesting mozilla project drawing on datomic / datascript : https://github.com/mozilla/mentat — originally clojurescript now rewriting in Rust.